More Than 20 Detailed Case Studies from Bloomberg Businessweek

Real Time Research REPORTS

Real-Time Enterprise Stories

• Adobe Systems • eBay • Maple Leaf Foods • Alliander • EMC • Mercedes-AMG • ARI • Florida Crystals • NFL • CareFusion • Globus • Nomura Research Institute • City of Boston, Massachusetts • Hewlett-Packard • Norwegian Cruise Line • City of Cape Town, South Africa • HSE24 • Southern California Edison • Commonwealth Bank of Australia • Johnsonville Sausage • T-Mobile • ConAgra • Kaeser Compressors

More than 20 detailed case studies from Bloomberg Businessweek Research Services and Forbes Insights featuring leading-edge enterprises across industries to explore the real value of the in-memory platform: SAP HANA. Sponsored by:

October 2014 Real-Time Enterprises A Research Collection

Contents

Forward: Making Decisions in the Moment...... 5

Adobe Systems: “Adobe Gets Personal With Customers” (high tech)...... 6

Alliander: “Alliander Energizes Its Business with Real-Time Analytics” (utilities)...... 10

ARI: “In-Memory Computing Drives Fleet Savings at ARI” (automotive)...... 14

CareFusion: “Real-Time Business Leads to Healthy Performance at CareFusion” (healthcare) ...... 18

City of Boston, Massachusetts: “Boston’s Better Community Connections Through Big Data and Analytics” (public sector) ...... 22

City of Cape Town, South Africa: “Real-Time Data Speeds Services in ‘The Mother City’ of Cape Town” (public sector) ...... 25

Commonwealth Bank of Australia: “CBA Offers More Personalized Banking Through Big Data and Analytics” (banking)...... 28

ConAgra: “Real-Time Business Leads ConAgra to Profitable Insights” (consumer products)...... 31

eBay: “Big Benefits from Big Data at eBay” (high tech)...... 35

Page 2 Real-Time Enterprises A Research Collection

Contents

EMC: “The Key to EMC’s Technology Growth: Integrated Acquisitions” (high tech)...... 39

Florida Crystals: “Real-Time Business Sweetens Performance at Florida Crystals” (consumer products)...... 44

Globus: “Globus Is Always in Style with In-Memory Technology” (retail).. 48

Hewlett-Packard: “HP Paves the Future with Real-Time Insights” (high tech)...... 51

HSE24: “Using Real-Time Insights, HSE24 Gets Closer to Customers” (retail)...... 55

Johnsonville Sausage: “Johnsonville Cooks Up a Real-Time Recipe” (consumer products)...... 59

Kaeser Kompressoren: “Kaeser Puts Customers First with Big Data and Real-Time Business” (industrial machinery & components)...... 63

Maple Leaf Foods: “Maple Leaf Foods Turns Over a New Leaf with Up-to-the-Minute Insights” (consumer products)...... 67

Mercedes-AMG: “Mercedes-AMG: A Showcase for Real-Time Business Decisions” (automotive)...... 71

National Football League: “The NFL’s Advanced Analytics Score with Football Fans” (sports & entertainment) ...... 75

Page 3 Real-Time Enterprises A Research Collection

Contents

Nomura Research Institute: “Tokyo Drivers Avoid Traffic Jams with NRI’s Big Data and Analytics” (high tech)...... 79

Norwegian Cruise Line: “Norwegian Cruise Line Sets Its Course for Real-Time Insights” (travel & transportation)...... 82

Southern California Edison: “How Southern California Edison Harnesses Big Data’s Power” (utilities)...... 86

T-Mobile: “Using Social Media to Improve Customer Loyalty at T-Mobile” (telecommunications)...... 90

COPYRIGHT AND DISCLAIMER NOTES Real Time Research Reports, a subsidiary of Triangle Publishing Services Co., does not make any guarantees or warranties as to the accuracy or completeness of this report. Real Time Research Reports shall not be liable to the user or anyone else for any inaccuracy, error or omission, regardless of cause, or for any damages resulting therefrom. In no event will Real Time Research Reports nor other companies or third-party licensors be liable for any indirect, special or consequential damages, including but not limited to lost time, lost money, lost profits or lost good will, whether in contract, tort, strict liability or otherwise, and whether or not such damages are foreseen or unforeseen with respect to any use of this document. This document, or any portion thereof, may not be reproduced, transmitted, introduced into a retrieval system or distributed without the written consent of the copyright owner.

The names of actual companies, publications and products mentioned herein may be the trademarks of their respective owners.

ELECTRONIC VERSION AVAILABLE To see or use an electronic copy of this document in PDF format, please visit the following Web site: www.saphana.com

Page 4 Real-Time Enterprises A Research Collection Forward: Making Decisions in the Moment

Many different factors impact the decisions companies make everyday. Being a real-time business means being able to make those decisions when they matter most. For most companies today, however, batch processes must run before any decision can be made. In retail, for example, personalized offers in stores aren’t common. Instead, stores offer a generic 25 percent off. But what if something personal and relevant for individual shoppers could be delivered when they walk in the door?

There will be a time when Real-time planning, for example, is the “game we walk into a store and our changer” for one SAP customer. Using HANA experience will be radically has reduced the time it takes to process one different. That’s why the SAP financial report from 40 hours to 20 seconds— HANA platform exists: to a 7,000 times improvement. enable companies to do business in the moment. I envision the SAP HANA platform not only Steve Lucas, enabling customers to make today’s decisions President For large companies, three but also providing them a powerful predictive Platform main drivers deliver value. engine for tomorrow’s choices. Most compa- Solutions, First, simplify: Reduce the nies make decisions by looking in the rearview SAP complexity of the systems mirror, but a rearview mirror offers only a tiny required to produce existing results. In fact, piece of the puzzle versus the forward view massive IT simplification was one of the key a windshield provides. Eventually, operating design motivations behind SAP HANA. You models will include built-in forward-looking can use the platform to feed data from all decisions. different sources into one system. Second, enable agility: Give customers insights in real Real-time business transformations are not time to aid decision-making. Third, innovate: just about technology, though. SAP HANA is Unlock the true potential of real-time business extraordinarily innovative, but the first thing innovation through new business processes we do is look at where the opportunities are and models. to transform business processes. Then, SAP helps customers rethink how those processes SAP HANA deployments began more than are designed and how they can be remodeled. three years ago. Since then, companies have You have to start at zero. If you don’t have to added the SAP Business Suite powered by wait for information, there’s an opportunity for HANA. Now there are more than 4,000 SAP massive reinvention and value creation across HANA customers. So far, business transforma- industries. n —Steve Lucas tions have been process by process based on the time it takes to get data and then analyze it. When you eliminate the latency, you have to think through how you want to work differently.

Page 5 Real-Time Enterprises A Research Collection

Adobe Systems at a glance Producer of desktop publishing and graphics editing programs as well as a wide array of products made for creative and marketing professionals

Industry: high tech

Founded: 1982

Headquarters: San Jose, Calif.

2013 revenue: $4.06 billion

Employees: More than 11,000 worldwide

Products: Software for creative and marketing professionals

www.adobe.com

Source: Adobe Systems

Page 6 RESEARCH UPDATE | Real-Time Enterprise Case Studies

Adobe Gets Personal With Customers With in-memory computing and predictive analytics tools, the high-tech company is anticipating customer needs and developing more personalized programs. BY JOE MULLICH hroughout high-tech, long-time business models are falling by the wayside as Tproducts become digitized, customer demands rise and the need for speed increases. Few companies know this better than Adobe Systems, best known for its Adobe Systems at desktop publishing and graphics editing programs Acrobat and Photoshop, as well as a Glance a wide array of products made for creative and marketing professionals. } Founded: 1982 } Corporate headquarters: Over the past few years, Adobe’s business model FIGURE 1 Changes to Adobe’s San Jose, Calif. has fundamentally changed. Instead of selling } Revenue: $4.06 billion, Business Model boxed software, its business is now driven by The company’s business model has fiscal 2013 fundamentally changed. } Employees: More than cloud-based subscriptions. And instead of products 11,000 worldwide being released every 18 months, new offerings and Past Present } Products: Software for upgrades are a constant. The company’s customer Offerings Sold by the Sold by the creative and marketing interactions—once periodic and performed through box subscription professionals resellers or partners—are now continuous and Distribution Sold through Sold directly

occur across multiple channels, including social partners www.adobe.com Source: Adobe Systems media, display ads, e-mail, the call center, direct Product 18-month Continuously releases cycles updated sales and the Web (see Figure 1, “Changes to

Adobe’s Business Model”). Source:Customer Adobe Systems Periodic Ongoing interactions relationship

These shifts have transformed the types of Source: Adobe Systems customer, product and sales data to which Adobe has access. Whereas customer information used to the exact data they needed rather than sorting be limited to name, address and billing information, through data dumps. To accomplish both of these Adobe now collects data on how customers use goals, Adobe needed to enable cross-functional its products, and in what context. Used correctly, collaboration, integrate silos of data, deliver one such data can lead to a better understanding of version of performance metrics, react to customer customer behavior and even future needs. signals in milliseconds and enable data-driven action.

The problem was, while each of its engagement According to Prasad Bhandarkar, director channels provided customer insights, the views of Adobe Information Services, these goals were fractured and channel-specific. Adobe needed were accomplished by leveraging in-memory to gain a more holistic view of customers, as well computing,1 which rapidly aggregates and analyzes as real-time insights into their behavior to deliver a vast amounts of numerous types of data. The highly personalized, engaging experience. It had use of in-memory computing and analytics tools also become essential for the company to quickly is the centerpiece of the company’s vision of obtain up-to-the-moment business performance “revolutionizing the way Adobe engages with 1. Prasad Bhandarkar spoke at a recent Webinar. information so that workers could quickly find information,” Bhandarkar said.

A Fractured Data Environment FIGURE 2 Channel Quagmire Adobe’s Photoshop software is currently installed on Respondents were asked, “How would you rate your ability to quantify the impact of one channel’s 600 million desktops, and 90 percent of the world’s performance on another (e.g., online display advertising’s creators use the graphics program. The company impact on search marketing)?” also offers Creative Cloud, which enables access to I don’t know Very good Adobe applications, and Adobe Marketing Cloud, 7.7% 4.8% which provides analytics and other marketing tools.

Adobe’s Business Good Challenges Adobe faced many of the issues common to Poor 16.8% 18.8% } Transforming into a technology businesses grappling with this new age cloud business of producing and marketing software. Although the } Using real-time data Fair company could see how a customer responded to a to tailor personalized 51% customer experiences specific marketing initiative, for instance, it could not } Creating company-wide easily combine that information with that customer’s

performance KPIs and e-mail messages and social media sentiment. A *Total does not equal 100% due to rounding

holistic customer views 2 recent study found that most marketers do not Source: The CMO Club } Enabling consistent have a holistic view across channels (see Figure 2, messaging across “Channel Quagmire”). and dynamic access to data. Real-time data numerous marketing channels acquisition puts instant decision-making at the } Aggregating structured Further, performance reports were often fingertips of people throughout the organization. and unstructured data inaccurate and inconsistent. Analytics was highly • Versatility. The platform enables employees to from multiple sources fragmented across the company, with different visualize data using a wide variety of charts and groups producing different numbers. And because graphs, so they can make connections in the executives relied on IT to query data, they had to data more easily and share their insights readily wait for results. The time delay was increasingly throughout the enterprise. unacceptable for an Internet-based business in which second-by-second customer activity can To ensure that the dashboard optimizes influence marketing and sales efforts. decision-making, the team drew inspiration for its visualization techniques from the Louvre in Developing a Dashboard Paris. “There is art and poetry to make this work,” A key piece of Adobe’s data strategy is Adobe Bhandarkar said. Users now get answers and Dash, a business intelligence platform that connects insights rather than tables and numbers, enabling data across multiple sources in real time. Because them to discover something new about the traditional databases cannot aggregate data on the business every time they use the system. fly, the company turned to in-memory computing. Adobe had used in-memory computing in the past to Presenting data effectively is a key component of real- analyze large data sets to stem software piracy,3 and time analytics. In a 2013 study by TDWI Research, the effort had quickly uncovered enough information respondents named several business benefits of 2. The CMO Club and Visual IQ. “Building Bridges to the Promised about software misuse to suggest that the technology data visualization technologies, including improved Land: Big Data, Attribution offered significant revenue opportunities. Adobe Dash operational efficiency (77 percent), faster response to & Omni-Channel—A CMO Perspective.” 2014 offers several benefits: business change (62 percent) and the ability to identify http://goo.gl/DaItYQ 4 3. SAP. “Adobe: HANA Customer • Trust. Business performance KPIs are now timely, new business opportunities (59 percent). Testimonial Video.” http://goo.gl/LD5Yxx accurate and undisputed. With a single set of 4. Stodder, David. “Data KPIs, Adobe can now work from a single version New Business Insights Visualization and Discovery for Better Business Decisions.” TDWI of the truth. With the ability to quickly aggregate information Research, 2013. http://goo.gl/xBYyLi • Speed. Business leaders now have direct from all channels via in-memory computing, Adobe

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

FIGURE 3 Preference for promotions. In the past, it was necessary to wait Personalization 24 hours to ascertain a promotion’s impact. With Consumers increasingly expect in-memory computing, that data is available within personalized interactions. minutes, enabling Adobe to tweak the promotion (Percent of respondents) immediately to improve results. Knowing what is Would trust businesses more if they explained how they are using personal information to improve happening with promotions within a 15-minute their online experience window is far more powerful than having a one-day Adobe’s IT 77% delay, and the agility for executive decision-making is game changing, according to the company. Solutions Get frustrated with Web sites that display content, } Fully integrated and real- offers, ads, promotions, etc., that have nothing to time customer profiles do with their interests The best forward-looking measure for any cloud across channels 74% business is how actively customers are using } Easy-to-use dashboard the product. For example, if a customer’s usage with a variety of Are OK with providing personal information on a of the Marketing Cloud or Creative Cloud starts visualization tools Web site as long as it is for their benefit and being } In-memory computing used in responsible ways to decrease a few months before the annual for instantaneous data 57% subscription ends, that is a clear warning sign,

aggregation and analysis Source: Janrain/Harris Interactive, 2013 and the sooner that Adobe can see this trend happening and respond to it, the better chance it can now provide more personalized customer has of retaining customers. interactions and a consistent experience across channels. For instance, when customers click on Adobe executives can easily see everything in the e-mails or display ads, the company can create an digital pipeline, including top customer issues, and identity composite for them, and tailor the message for further information, they can drill-down on the based on that information, as opposed to providing fly. The company plans to enable this capability on them with generic messaging. Such personalization mobile devices in the future, as well. In fiscal 2013, aligns with customer preferences; in a recent Adobe saw an uptick for its efforts. Creative Cloud study, consumers said they prized personalized subscriptions grew by 1.1 million, and subscriptions marketing that catered to their interests (see Figure to Document Services doubled to more than 1.6 3, “Preference for Personalization”). million. Adobe Marketing Cloud achieved a record $1.02 billion in annual revenue, representing 26 On the business performance side, when percent year-over-year growth. Adobe employees have questions about order management and booking, they can slice and dice Using in-memory computing, data visualization data in many ways—such as by financial quarter and advanced analytics, Adobe is able to fully or certain geographies—and receive an answer operate as a cloud-based company. It can see a in three seconds. “That capability never existed complete and updated view of customers across before,” Bhandarkar said. “As our business changed channels, personalize customer interactions and to a subscription model, it’s become even more make profitable business decisions based on in- important for us to know on a day-to-day basis the-moment data. By leveraging real-time big data what customers are doing inside the product.” analytics, high-tech companies such as Adobe are With Adobe’s in-memory system, an order that laying the foundation needed to compete today and was booked five seconds before shows up in the in the future. • aggregated numbers. Joe Mullich is a freelance business and technology writer based in Sherman Oaks, Calif. Additionally, Adobe can quickly analyze customer acquisition and usage patterns to optimize This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Alliander at a glance A regional grid operator for gas and electricity in the Netherlands Industry: utilities

Headquarters: Arnhem, Netherlands

Business units: Alliander manages the gas and electricity grids in many areas of the Neth- erlands; Liandon works on energy infrastructures for high-voltage, complex medium-voltage and industrial installations

2013 Revenues: €1,744 million

Employees: 6,000 www.alliander.com

Source: Alliander

Page 10 RESEARCH UPDATE | Real-Time Business Case Studies

Alliander Energizes Its Business with Real-Time Analytics A Netherlands utility is using real-time business insights based on big data to boost reliability, lower costs and transform customer relationships.

BY JOE MULLICH lliander, a regional grid operator for gas and electricity in the Netherlands, is Agrappling with an issue that is becoming widespread among utilities: unpredictable demand due to rapid changes in customers’ energy consumption Alliander at a behavior. Energy-hungry devices such as plug-in electric vehicles (PEVs), as well as Glance Headquarters: Arnhem, unconventional energy sources such as solar and FIGURE 1 Wide Range of Analytics Value Netherlands wind, have fundamentally changed the energy Respondents were asked in which areas they Alliander see analytics having the greatest impact on smart solutions Business units: landscape. Case in point: There are now over manages the gas and deployment. (% responding) 1,030 public charging points for PEVs in the electricity grids in many Grid operations areas of the Netherlands; regions where Alliander is active, of which more 96% Liandon works on than 300 were realized just last year.1 energy infrastructures for Asset management high-voltage, complex Such changes are heightening the need for Alliander 92% medium-voltage and to improve peak load forecasting in its networks industrial installations Outage management quickly and accurately. Failing to do so could lead to 2013 Revenues: €1,744 85% million (US$2,386 million) higher operating costs or, worse, network outages Employees: 6,000 that leave customers angry and without power. AMI operations www.alliander.com 77% Source: Alliander Consequently, a good part of Alliander’s future Demand response hinges on using real-time data to make 73% “unpredictable demand” more predictable by leveraging the information that smart meters, smart Customer operations grids and enhanced customer relationship 73%

management systems provide. The utility is applying Revenue protection/theft reduction this great mass of data to other real-time uses, as 52% well. Jill Feblowitz, vice president at consultancy IDC Base: 54 utilities in 13 countries Energy Insights, said that as utilities work to realize Source: Accenture, 2013. http://goo.gl/mwjiLS more value from their smart meter implementations, “analytics will be the trend that has the most impact Lots of Data, Big Payoff over the next five years.” In the utility industry, lots of data is accumulating rapidly: Smart meters in consumer homes gather usage data; Indeed, a 2013 Accenture survey revealed a broad the smart grid collects data from the devices in the array of business functions that utilities say would network; and a significant amount of data comes 2 1. Alliander. “Alliander is working benefit from better analytics (see Figure 1, “Wide courtesy of those who generate energy with solar on a sustainable future.” Alliander Range of Analytics Value”). The need to use panels or wind power. And utilities expect a big return press release, Feb. 14, 2014. 2. Accenture. “Unlocking the Value dynamic and granular data to make decisions in the from all this data. A 2013 survey by Tata Consultancy of Analytics.” Research report, 2013. http://goo.gl/mwjiLS. moment is fast becoming an industry imperative. Services found that both utilities and energy/resources

companies have the highest expectations for FIGURE 2 Big Data Expectations 3 generating returns on investment in big data (see Of all industries, utilities expects the biggest Figure 2, “Big Data Expectations”). returns on its big data investment. (mean percentage of expected ROI for 2012 big data investments)

Alliander itself has been on a data quest. The utility Utilities dramatically increased the number of sensors 73.0% Alliander’s throughout its grid, enabling it to gather larger data Energy/resources Business volumes for analysis. Measurement data is sent at 60.6% Challenges five-minute intervals, and the sensors generate 3.15 High tech } Improve forecasting of billion records every year—a huge repository to mine 52.4% peak energy usage. for information and insights.4 } Provide customers with Average greater insight into 45.5% energy consumption. To analyze the data quickly enough to apply insights } Gain fast access to large in a timely fashion, Alliander turned to in-memory Banking/financial services 43.7% volumes of computing, technology that enables very large measurement data for volumes of diverse sources of data to be rapidly and Insurance analysis. easily analyzed. The utility has seen a number of 38.9%

Alliander’s IT benefits, including more accurate forecasting of Travel/hospitality/airlines Solutions energy demand, greater efficiency by automating 37.9% } Expanded network to manual tasks, improved auditing and reduced Telecommunications 22,000 sensors spread 5 energy costs for customers. 37.9% across 400 substations. } Implemented in-memory Transforming Customer Relations Source: Tata Consultancy Services, 2013 computing to speed data analysis. According to a 2013 IDC Energy Insights study, the } Set up wireless mobile highest business priority for utilities today is meters send granular data on energy consumption directly telecommunications increasing customer satisfaction, followed closely by to utility companies, and some utilities use small wireless network. reliable service delivery and operational cost displays to show people how much energy and money reductions.6 Analytics helps utilities achieve this goal they are spending in real time. A survey by uSwitch, an by changing the customer relationship to a energy price comparison service in the United Kingdom, partnership. For instance, using smart meter data, found that consumers with smart meters are more likely utilities can work with customers on using energy to partake in energy-saving behaviors8 (see Figure 3, more efficiently and reducing their bills. “Satisfaction with Smart Meters”).

Customers can also play a role in helping utilities At Alliander, customers who have real-time access to achieve their own conservation goals. Robin data about their energy usage have reduced energy bills 3. Tata Consultancy Services. Hagemans, manager of grid information and control at by 10 to 20 percent per month. “We need to help our “The Emerging Big Returns from Big Data.” Research report, 2013. Alliander, noted the utility’s aggressive goals of reducing customers at the household level to use their energy http://goo.gl/NhEUgW. 4. Tekurkar, Shailesh. “Leveraging carbon emissions by 20 percent by 2020 and boosting much more wisely,” said Jeroen Scheer, manager the ‘Smart Grid’ for Smart 9 Decisions on Network Asset renewable energy use by the same amount. That taskforce energy transition IT at Alliander. Replacements.” Analytics from SAP blog, Nov. 13, 2012. goal, he said, can be reached only if consumers http://goo.gl/E5PD6V. become more active participants in the process of Forecasting Peak Loads 5. Shailesh Tekurkar. 6. Feblowitz, Jill. “The Digital conserving energy and balancing their energy use.7 Perhaps an even bigger advantage of analyzing data from Utility.” IDC Energy Insights executive summary. September smart grids and smart meters in real time is the capability 2013. to accurately forecast demand, which enables utilities to 7. Koster, Stefan. “How Alliander The rapid speed of in-memory computing produces is accelerating its load forecasting the insights needed to do just that, and it helps predict and head off likely outages, as well as to identify process using SAP Analytics.” SAP presentation. May 28, 2013. Alliander better engage with its customers. Smart leaks and fraud. Better forecasting enables utilities to plan

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

future investments that prevent network overloads. FIGURE 3 Satisfaction with Smart Meters Such forecasts are tricky, though, because the Customers with smart meters report high substation load depends on numerous variables— satisfaction with their provider. (% responding) demographic growth, seasonality, consumption Are happy with their smart meter forecasts, specific events and more. 92%

Would recommend having a smart meter installed In the past, collecting and analyzing the enormous Real-Time 87% Business Benefits amounts of data required 10 weeks of effort. As a at Alliander result, Alliander would only perform forecasting and Use it to monitor energy usage and reduce } Reduced time for grid optimization once a year. With in-memory consumption forecasting and grid 81% computing, that process can be accomplished in a optimization from 10 mere three days. Consequently, the utility can weeks to three days. Are more likely to turn off lights when not in use } Implemented 85 new monitor the transformer stations, detect sensor 51% business models to problems automatically and forecast future power Better understanding of which actions consume analyze operations. loads once a month. This provides a more the most energy } Helped customers accurate forecast, much deeper insight and greater 41% reduce monthly energy opportunities to drive efficiency.10 bills by 10 to 20 percent. Base: 5,057 U.K. households with a smart meter installed Source: uSwitch, 2013 “We can optimize our grid and create new models we couldn’t think of a year ago,” Scheer said. Such efforts include setting up a wireless mobile Indeed, as a result of the efforts, Alliander has telecommunications network to facilitate the exchange created 85 new models to analyze its operations of all information within the utility’s grids. “The impact of more effectively. “We found out that 5 percent of the energy transition is becoming increasingly visible,” the peaks determined last year were inaccurate,” he said. “For instance, more and more renewable according to Stefan Koster, BI & analytics solution energy is being generated, and this will have major architect at the utility. The new algorithms help consequences for the networks in the long term.”14 detect more errors, leading to more efficiencies.11 In the IDC Energy Insights report, Feblowitz Business Reinvention recommended that utilities fortify their information

8. Martinelli, Michele. “New Market analyst firm GTM Research predicted and telecommunications infrastructure and research shows consumers are happy with smart meters.” USwitch global utility company expenditures on data and applications so they can support advanced analytics news, Dec. 12, 2013. analytics will grow from $700 million in 2012 to such as forecasting, predictive load, simulations and http://goo.gl/0UDnUi. 9. Scheer, Jeroen. “Alliander $3.8 billion in 2020, with gas, electricity and water optimization. One specific capability she cited is and SAP HANA.” SAP customer testimonial video. suppliers in all regions of the world increasing integrating demand information with geospatial http://goo.gl/cS0CoV. 10. Courtney, Martin. “How their investment. This will result in “a complete visualization tools. Alliander is already going down utilities are profiting from Big reinvention of the utility business,” according to this track.15 Data analytics.” Engineering and Technology Magazine, Jan. 20, analyst David J. Leeds.12 2014. http://goo.gl/r60uhi. 11. Stefan Koster. As Pieter den Hamer, manager of business intelligence 12. Martin Courtney. 13. Alliander press release. As Alliander looks to the future, it is seeking to boost and analytics at the utility, put it: In the current utility 14. Alliander press release. the insights gained from network data. Peter landscape, “Innovation is not just something extra, but 15. Hagemans, Robin. “Alliander Smart Grid Analytic Use Cases Molengraaf, Alliander CEO, said the company’s main it’s a necessity.”16 • Based on a Live Mid-voltage Grid Lab.” OSIsoft EMEA industry focus is “making the right choices for the future of session, 2013. Joe Mullich is a freelance business and technology writer http://goo.gl/ubKnOy. our networks. Whilst continuing to maintain and based in Sherman Oaks, Calif. 16. Szirtes, Tamas. “Business replace our existing electricity and gas networks, we Innovation Experience – a resounding success.” Intenzz blog, are also preparing for the radically changing energy This research project was funded by a grant from SAP. Feb. 6, 2013. 13 http://goo.gl/O9D3Wc. landscape of the future.”

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

ARI at a glance Largest privately held fleet management services company in the world

Industry: automotive

Parent company: Holman Automotive Group

Employees: 2,400 worldwide

Headquarters: Mount Laurel, New Jersey

www.arifleet.com

Source: ARI

Page 14 RESEARCH UPDATE | Real-Time Business Case Studies

In-Memory Computing Drives Fleet Savings at ARI The world’s largest privately held vehicle management company is turning big data into real-time insights and predictions that deliver unprecedented business value to customers. BY ESTHER SHEIN WISHNOW

n employee driving a company car or truck may think he is doing the business a big A favor by purchasing gas at a low-cost fuel station a few miles out of his way. What he cannot know is that the station has a higher incidence rate of accidents because it is located

at a dangerous intersection. But ARI knows. Thanks to can be made in minutes, not months. With its data

AN EXCLUSIVE in-memory technology, the company is quickly turning volume doubling every 14 months, that has become an RESEARCH REPORT seemingly unrelated datapoints into insights its customers ever-greater challenge, which is why ARI invested in an FROM BLOOMBERG can use to optimize fleet efficiency—before they make a in-memory database. The solution has reduced response BUSINESSWEEK wrong turn. times on delivering information to customers and solved RESEARCH SERVICES its “data latency problem,” says Tony Candeloro, vice ARI manages more than 950,000 vehicles in North president of product development (see Figure 1, “In- America and the United Kingdom. With its two million Memory Computing: The Payoff for ARI”). AT A GLANCE worldwide associates, the company collects up to 14,000 datapoints on each vehicle, including specifications, Real-time analysis of very large date sets is possible with ARI at a Glance maintenance, fuel, safety and value. It also captures high-performance, in-memory database technology, says Industry: telematics information through onboard computers, Hyoun Park, principal analyst at Nucleus Research. HANA Vehicle fleet management including how fast a vehicle is moving, when it makes a from SAP, for example, provides orders of magnitude Company description: left- or right-hand turn and where it has stopped. performance increases by accelerating information delivery Largest privately held fleet and queries, he says. “A lot of processes that have slowed management services In the past, ARI used this information to provide reports down as companies deal with huge amounts of data company in the world and recommendations to customers. But staying are now being brought back to operational levels where Parent company: competitive today means delivering insights so decisions companies can get data quickly again.” Holman Automotive Group Employees: FIGURE 1 2,400 worldwide In-Memory Computing: The Payoff for ARI Headquarters: Running a report or query before in-memory Up to two hours; some reports were so complicated Mount Laurel, N.J. technology they would time-out after 24 hours Web site: www.arifleet.com

Running a report or query with in-memory technology Three seconds

Number of datapoints captured per vehicle 14,000 (late 2011 into early 2012)

Number of vehicles in ARI’s fleet Roughly 950,000 in North America

ROI 5% efficiency improvement in the call center

Source: ARI

Speed Equals Value Moving those large silos into a real-time environment was This was the situation at ARI: the need to drive the a challenge, says Bob White, senior vice president of information it captured to its fullest potential—quickly. client and fleet relations at ARI. “Before we implemented “Once you have all the data, it’s about how you analyze this technology, working with the data on a real-time and correlate it to make better decisions,” Candeloro basis to draw correlations and spot trends was difficult,” says. “While we have always helped our clients manage he says. The in-memory system provides employees ARI’s Data their fleets efficiently and worked to deliver the lowest with new real-time insights to create more value for Challenges total cost of ownership possible, the time it took to customers. “[We can] leverage that data in a timeframe Manage data growth: decipher all of the data and make day-to-day decisions that meets the business need,” White says. With 14,000 datapoints was becoming more difficult as our data volumes grew.” collected per vehicle, ARI’s Now, if a technician is told that a brake fix on a truck data volumes are doubling For example, to manage the cost of vehicle repairs, will cost $500, he can see 25 similar repairs on similar every 14 months. ARI’s technicians need to authorize the right repair at the vehicles within an identified geographical radius. This Speed decision-making: right cost. “Historically, a vendor would call us seeking information makes negotiating on the fly a lot easier. Data volumes were authorization for a repair. If our techs agreed with the Before, “we were doing millions of transactions, so we increasing the time it took to diagnosis, they would negotiate the best possible price didn’t even attempt it,” White says. When customers run reports and queries. for that repair by using their own expertise and the best needed more granular data, it required a lot of manual Improve insights: information they had on hand at that time,” Candeloro says. work. “Now, we can leverage more data to make better ARI wanted to arm fleet decisions,” he says, “and we’ve also improved the overall managers with real-time As time went on, he says, ARI wanted a solution transaction time by a little more than five percent.” insights so they could make that would enable better and quicker decisions. The decisions that optimized company implemented a data warehousing strategy A majority of ARI’s employees are using the system on efficiency. that ported data from its transactional database into a a daily basis, and 80 customers are using it through real-time in-memory database, built some data models the customer portal, ARI insightsTM. When customers and conducted a pilot with one of its customers. Then, submitted inquiries in the past, employees would search with the help of HP, ARI moved the system in-house in for information in the various systems and create reports, December 2011. By the start of 2012, Candeloro says, which could take several days to produce. “Now, our ARI was up and running, building queries and “data frontline employees can run the queries themselves,” universes” based on detailed data previously segmented White says. “By putting the tool in a customer’s hands, into separate silos on disparate systems. we are empowering our clients to make better decisions.”

FIGURE 2 ARI’s In-Memory Technology: What the Future Holds

Capability Example

Predictive data mining: Spotting trends and patterns Recommending a repair before a failure occurs. ahead of time to be proactive rather than reactive.

Benchmarking: Designing models that gather trend data Calculating a cost-of-ownership baseline across on all vehicles to create a baseline and then identifying fleets with similar vehicles and identifying those that outliers. exceed the baseline.

Changing driver behavior: Finding correlations between Discovering a link between fueling locations and events that seem unrelated at first glance and using that accidents. information to make recommendations.

Source: ARI

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

ARI’s salespeople can also work more effectively with FIGURE 3 customers and prospects, he adds. And in its call centers, which include about 400 employees, ARI has A Surge in Predictive Analytics increased efficiencies by about 5 percent, White says. Primarily using predictive analytics “That one area alone has probably more than covered our 12% investment, and we’ve just begun to discover all of the 33% 2009 Elements of ways to leverage it,” he says. 2012 ARI’s Real-Time Primarily using retrospective analytics 31% Analytics Solution Looking Ahead 29% In-memory database: Candeloro says the company has only skimmed the Using both equally Using an in-memory surface in terms of realizing the full potential of in- 50% database, ARI can quickly turn memory technology (see Figure 2, “ARI’s In-Memory 28% data into insights that help Technology: What the Future Holds”). ARI is planning its customers optimize fleet to use data mining capabilities to spot trends and Don’t know efficiency. patterns “so we don’t just react to a phone call but 1% 10% Integrated data: get ahead of it and use the predictive angle,” he says. Data “universes,” created For example, White says, employees have already Base: 600 companies in the United States and United Kingdom within the in-memory detected trends that predict needed repairs on a truck with 1,000-plus employees Source: Accenture database, integrate data from before a major failure occurs. “We can recommend to disparate systems. our customer that they preemptively have that repair Customer portal: done,” he says. station example. “If, for example, you know a driver A majority of ARI’s employees fuels at a particular gas station, and you also know this use the system daily, and Another goal is to begin benchmarking, building models group of vehicles has a higher percentage of accidents, select customers run their to look at all similar vehicles managed by ARI and once you look at those two together, you may make a own queries using the determining their baseline cost. This way, the company recommendation not to fuel at that location depending on customer portal. can easily identify the outliers and take corrective action, what the data actually reveals,” he says. Data mining: Candeloro says. ARI plans to use predictive According to a 2013 study by Accenture, more than analytics to spot trends and Keeping up with all of the variables involved in the cost twice as many respondents are now using analytics patterns among seemingly of fleet management—where vehicles are used, who is primarily as a predictive tool than in 2009 (see Figure 3, unrelated data points. driving them, the environment in which they operate, the “A Surge in Predictive Analytics”). “This surge reflects time of day, accidents, driving patterns and the amount a growing sophistication in analytics capabilities that of idle running time—is ARI’s main challenge. “We try anticipate tomorrow rather than explain yesterday,” to track and manage those variables, but a lot of them according to Accenture. interact with and impact each other,” Candeloro notes. “If you’re idling a vehicle a lot, you may need to do White says ARI is continuously thinking about whether the preventive maintenance more often. And because we data it is collecting can predict an outcome. “The ability have the ability to look at that information, we can push has always been there,” he says. Until now, “we just out recommendations to our customers. That was very didn’t have the tools to unlock it.” • difficult to do in the past.” Esther Shein Wishnow is a freelance writer and editor whose work has appeared in several online and print As the company moves further into predictive analytics, publications including Computerworld, Information- Candeloro says ARI hopes to find the not-so-obvious Week, BYTE, Network Computing and CIO. correlations between seemingly unrelated events, such as how fuel patterns impact accidents, as in the gas This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

CareFusion at a glance Makes products that help reduce medication errors and reduce health-associated infections

Industry: healthcare

Headquarters: San Diego, California

Customers: more than 25,000 worldwide, including hospitals, surgery centers, long-term care facilities, outpatient and ambulatory clinics, governments and insurance providers

Focus areas: medication management; infection prevention and surveillance; operating room effectiveness; device connectivity; data and analytics; respiratory care

Revenues: $3.55 billion (fiscal 2013)

Employees: more than 14,000

www.carefusion.com

Source: CareFusion

Page 18 RESEARCH UPDATE | Real-Time Business Case Studies

Real-Time Business Leads to Healthy Performance at CareFusion CareFusion employs in-memory computing to transform its business, improve operating margins and accelerate revenue growth.

BY JOE MULLICH

round the globe, demands for healthcare reform are causing the life sciences CareFusion at a Glance Aindustry and its customers, the care providers, to look for efficiencies. CareFusion, Headquarters: the $3.5 billion maker of products that help reduce medication errors and reduce health- San Diego, Calif. Customers: More than associated infections, intends to be at the front line 25,000 worldwide, FIGURE 1 Meeting the Challenge to Change of this transformation, promising quality and safety Life sciences CEOs view data and analytics including hospitals, among the initiatives that are critical to transformation. surgery centers, long- at reasonable cost. (% responding) term care facilities, outpatient and ambulatory In its efforts to achieve this goal, CareFusion has Channels to market 57% clinics, governments and averaged one major acquisition annually since its insurance providers spinoff from Cardinal Health in 2009. To date, the Use and management of data and data analytics Focus areas: Medication acquisitions have bolstered CareFusion’s product 56% management; infection line with technologies that improve respiratory care prevention and M&A strategies, joint ventures or strategic alliances surveillance; operating and deliver intravenous drugs to patients. But they 55% room effectiveness; have also created management challenges. device connectivity; data Customer growth and retention strategies and analytics; respiratory 50% Each of the acquired companies had its own care systems for strategic planning, which made it Approach to managing risk Revenues: $3.55 billion 50% (fiscal 2013) difficult for CareFusion to get a global view of its data. Yet comprehensive planning was essential for Employees: More than Source: PricewaterhouseCoopers, 2014. http://goo.gl/QgoZwG 14,000 the company to reach its aggressive goals. “The first www.carefusion.com two years of growth we called a ‘standup’ phase, CareFusion found a way to assemble a global view of its

Source: CareFusion which was the separation” from Cardinal Health, operations to improve the collection and consolidation of CareFusion CEO Kieran T. Gallahue told analysts in information it needs for better planning and strategic 2013. “The last two years began what we call the decision-making. The company is using financial ‘building the foundations for growth’ stage,” which, forecasting and planning software in combination with he said, could only be achieved by streamlining the in-memory computing, a new database and analytics organization and reducing costs.1 platform that enables very large volumes of data to be aggregated and analyzed quickly to answer any question The same theme resonates throughout the life almost instantly. Organizations frequently use in-memory sciences industry. An annual global survey of technology to make more accurate budget forecasts, run pharmaceutical and life sciences CEOs by “what-if” scenarios faster and gain better insight into PricewaterhouseCoopers found transformation on customer relationships.

1. Nasdaq information on their corporate agendas, including new business CareFusion. http://goo.gl/1Ncrfb. structures, technology investments, and the use of “The entire healthcare industry has been inefficient 2. PricewaterhouseCoopers Global CEO Survey: Pharmaceuticals and data and analytics2 (see Figure 1, “Meeting the relative to other industries,” observed Mike Zill, executive Life Sciences. 2014. http://goo.gl/QgoZwG. Challenge to Change”). vice president and CIO at CareFusion. “That needs to

change, and we are dedicated to being a force that FIGURE 2 CareFusion’s Complex Operations helps lead the change.” The company needs to consolidate and analyze data about two dozen major brands in four product areas.

A Unified View Product area Number of brands More than 14,000 CareFusion employees around the world manufacture products purchased by more than Infection prevention 4 CareFusion’s 25,000 healthcare providers. That makes the Respiratory care 6 Challenges company’s operations complex to manage, Zill } An acquisition spree explained (see Figure 2, “CareFusion’s Complex Medication management 8 resulted in disparate Operations”). Meanwhile, the pace of business ERP systems that could Operating room effectiveness 6 demands that CareFusion be able to run multiple not share data. } Financial, employee business scenarios quickly, using the most accurate Source: CareFusion. http://goo.gl/rtgQkn and customer data was data possible. difficult to consolidate and it would take to run one, Zill said. Current numbers analyze in a timely way. Yet data from each unit had to be pulled out manually. lead to better estimates, more precise analysis and } Executives lacked Only averaged, historical numbers were available for improved decisions. CareFusion can determine confidence in the analysis. As a result, reports were produced slowly, within minutes how departments are performing and accuracy of the data used to make decisions. and their reliability was hard to assess. “Data is a the potential impact of adjustments. Michael DeLeo, reflection of the organization,” said Wayne Eckerson, financial manager for financial systems at CareFusion’s director of the research, business applications and CareFusion, said the company’s efforts are in line Solutions architecture group at consultancy TechTarget. “If you with an industry trend to compile and consume data } Deployed real- have fragmented, autonomous business units, you’ll more rapidly.3 time planning and have fragmented data.” consolidation software that automates and More Timely Action streamlines planning, So the CareFusion finance team led the charge to What is more, because CareFusion executives can budgeting, forecasting unify the company using real-time analytics. “There is now accurately track, in real time, how their and consolidation a saying: Time kills all deals,” Zill continued. “If the adjustments play out, they can more quickly agree on activities. executives are having a discussion, they won’t accept specific actions that different groups need to take to } Adopted in-memory waiting while you pull together the data they need.” achieve financial goals. computing to speed data collection and A survey by IDG Enterprise found improving the analysis. quality of decision-making, making quicker decisions, When building a budget or forecast, for example, and improving planning and forecasting to be the top CareFusion often needs to model changes to business drivers for investing in analytics (see Figure employee-related costs, like benefits, tax rates and 3, “Better Decisions Through Data”). international currency changes. Previously, adding a new benefit or calculating the impact of a health During each fiscal year, CareFusion compiles and insurance cost increase required a lot of manual reviews several iterations of its budget, adjusting for work. Information would be gathered from local business results. Before the company was able to controllers and payroll groups to determine which quickly assemble and analyze its financial data, it employees would be affected. The financial was difficult to determine whether departments were department then would forecast the impact of the meeting their individual targets. Without this proposed changes and apply the anticipated cost for information, CareFusion could not make timely each group to its budget. course corrections. 3. “Controlling 2013 Speaker The process took a week or more, and verifying the Profile: Michael DeLeo, Finance Systems Manager at CareFusion, With real-time data, however, the company can run accuracy of the data was difficult. “There was no easy Inc.” ERP Corp. blog, 2013. http://goo.gl/PZGAZx. more—and more sophisticated—models in the time method to ensure that all required employees and cost

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

centers have been accounted for in the adjustment,” FIGURE 3 Better Decisions Through Data Zill said. If the actual costs turned out to be different Business drivers for investment in analytics than the forecast, it was equally challenging to include the following. (% responding) determine exactly why. The variance could be due Improving the quality of decision-making to one or more factors—including incorrect 59% calculation methods or changes in expected Making quicker decisions Benefits of Real- employee headcount or benefit eligibility. 53% Time Business at CareFusion Now, the company can make more precise Improving planning and forecasting 47% } Increased availability: forecasts at the enterprise level, and make them in Information to make near real time. And when variances between Developing new products, services and key decisions that took planned and actual costs occur, CareFusion can revenue streams more than a week to 47% pull together can now determine the reason more easily, according to Zill,

be delivered in minutes. and decide quickly the best way to bring the budget Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX } Deeper insight: back in line. That is essential to achieving the Analysts can explore efficiencies the life sciences industry seeks, he said. big payoff, both to expanding margins and the questions of profitability bottom line.4 Now that the company has proven that in detail at both the New Perspectives and Insights the in-memory platform and advanced analytics customer and the product level, as well Zill added that the most valuable benefit of having models enable it to understand its costs at a greater as run more “what-if” real-time data is obtaining new insights. For level of detail, with more accuracy and in less time, it scenarios. example, CareFusion can model profitability down to intends to expand these capabilities to other models } Improved forecasts: both product and customer levels, giving it a more and other data sources, Zill said. By using real-time data, granular view to aid decision-making. and gaining flexibility For example, he added, for consumable products, to include more data sources in its analyses, In the past, Zill said, the company might have only CareFusion plans to model sales based on historical CareFusion can been able to get a partial view of its relationship trends and seasonality. create more accurate with a customer. Now that it can consolidate and forecasting models. analyze all the data from different units, “These revenue models will provide us with a basis CareFusion will use this technology to discover, in for forecasting manufacturing costs and profit the moment, which units customers purchase margins,” Zill said. “They will allow us to further from and how best to allocate company resources improve our accuracy, efficiency and ability to plan to properly serve customers. for changing business needs.”

“We can approach the model-making at a much Zill envisions expanding the use of in-memory greater level of detail,” Zill explained. “This happens computing to many additional areas—including because we no longer have to worry about the size mergers and acquisitions, research and of the data anymore. In the old days, you had to development, sales and marketing—which will bring think in terms of the minimum amount of data you the enterprise-wide improvements that are needed could bring together and still answer the question. to propel CareFusion, and its healthcare customers, Now we don’t have to be so smart about eliminating to a new era of higher efficiency. • information, so we don’t throw away nuggets. That’s Joe Mullich is a freelance business and technology mind-freeing and time-freeing.” writer based in Sherman Oaks, Calif.

Consider More Data This research project was funded by a grant from SAP. According to CareFusion’s Gallahue, the intense 4. Nasdaq information on CareFusion. http://goo.gl/1Ncrfb. business integration work has already produced a

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

City of Boston, Massachusetts at a glance Capital of the State of Massachusetts, founded in 1630 Industry: public sector

Population: 626,000

Size: 48 square miles

City employees: 16,173

Education: 53 institutions of higher education

Firsts: U.S. public school, subway system www.cityofboston.gov

Source: City of Boston

Page 22 RESEARCH UPDATE | Real-Time Enterprise Case Studies

Boston’s Better Community Connec- tions Through Big Data and Analytics With in-memory computing and analytics tools, the City of Boston is providing better service to citizens and engaging more with the community. BY JOE MULLICH any government agencies talk about the transformational power of big data and analytics. MBill Oates, chief information officer for the City of Boston, is doing something about it. He says this starts with making better connections—between citizens and the community, multiple government departments and even other cities.

“Business process change is table stakes at this point,” Boston at a Glance Oates says. “The importance of the technology is its Citizens Connect: Empowering } Founded: 1630 ability to engage and empower our constituents.” Constituents with Big Data and } Population: 626,000 Analytics } 34% of citizen reports via mobile and online app } Size: 48 square miles For Boston, a new form of engagement began in 2008 } 89% of citizens would recommend it } City employees: 16,173 with the launch of Citizens Connect1, a system that } 21% rise in constituent satisfaction } Education: 53 institutions enables citizens to report potholes, graffiti, damaged } Pothole repair time cut in half of higher education signs and other issues through the Internet and, later, } Firsts: U.S. public their smartphones. A unique twist of the technology, school, subway system then and now, is that citizens do not simply report Like many cities, Boston has a growing number of sources www.cityofboston.gov/ the problems to the government. Rather, reports and of data to analyze. To help it cope, the city is ahead in Source: City of Boston photos are published anonymously online, spreading taking visual analytics to a new level. For example, in 2010 word of the issues and inviting discussion and the Boston Police Department opened a “real-time crime participation in Citizen Connect. center” that receives dozens of feeds from street cameras around the city. The resulting data gives researchers Since then, Boston has introduced Street Bump2, the potential to visually analyze and match videos from which enables people to use their smartphone’s incidents to help identify suspects, mobilize resources and accelerometer—a motion detector in the device—to even map evacuation routes during emergencies. record road conditions and send data to public works employees. Unlike Citizens Connect, the Street Bump Big Data Makes More Satisfied Citizens app does not require a citizen to take action to report Boston’s experience with in-memory computing and issues. He simply turns on the app and, as he drives, big data analytics reveals intriguing insights about data is automatically collected and sent to the city. the opportunity for collecting and using big data. For example, residents were polled on why they failed to Street Bump was expected to identify the location of contact the city about maintenance issues before the potholes—a top concern of Boston residents. Thanks to Citizens Connect app. Their answer: When they call the analytics, the early data has provided some unexpected city, they feel like they are complaining; when they use the insights: trouble spots are eight times more likely to be app, they feel like they are helping. “castings,” those manhole covers, grates and other cast- metal lids that are supposed to be flush with the roadway The ability to rapidly share data and analytics has had a

1. Citizens Connect: Making Boston surface but instead heave up due to the extreme cold of measurable effect. Today, 20 percent of citizen reports Beautiful. http://tinyurl.com/642ymk3 a New England winter. Hundreds of these castings have come through the Citizens Connect app and another 2. Street Bump. http://tinyurl.com/mjeb4q8 been repaired as a result. 14 percent through the city’s Web site. The number of

citizen reports has doubled, as well, from 40,000 a year prior to the technology to 80,000 a year today. A reason Boston’s IT Challenges for the greater participation may be the quick results that } Quicken innovation follow. Since the city implemented Citizens Connect, the } Encourage citizen involvement response time to fill a pothole has been cut from three } Share big data across departments } Speed repair of potholes and other issues days on average to half that amount. Boston’s IT Solutions Similarly, the original prototype of Street Bump could not “We met with interested } Formed the Office of New Urban Mechanics— distinguish between potholes, manhole covers, bridge researchers, technol- an incubator that pilots new ideas quickly surfaces and other obstacles, so the app generated a lot } Partnered with citizen groups, businesses and ogists and community of false positives in the field.3 A crowdsource challenge universities to share data and ideas members to figure out enabled the city to implement a new algorithm to analyze } Developed apps that enable citizens to report how we could use data the big data coming in more accurately. issues easily } Developed apps that enable citizens to collect data to improve quality of life about road conditions automatically as they drive In Oates’ view, big data is all about information sharing in the issues. Being able to public sector. He has found that the combination of the right connect newly available data, timely analysis and visualization is most effective when city data with some of information and analytics are shared across departments. It also indicates the new thinking that is needed to the real big data out “In the past, an agency might crunch its own data to leverage the power of big data. Oates points to the Office there in the world can get a historical comparison on how quickly it is fixing of New Urban Mechanics, an initiative that began in potholes,” he says. “Now, by combining data, we can see Boston and is now shared with the City of Philadelphia make city government what areas of the city aren’t getting phone calls, which and serves as each city’s innovation incubator. The more proactive—and may indicate problems no one knows about yet.” office builds partnerships between city agencies, outside more effective.” institutions and entrepreneurs to pilot projects in the two Boston has put significant effort into making data and cities that address resident and business needs.5 Several —BILL OATES, CHIEF INFORMATION OFFICER, CITY analytics available to both agencies and the community. projects have involved data sharing and analytics, such OF BOSTON Oates points to the Boston Area Research Initiative4 as as the Street Bump initiative. an example. Here, the city confers with the many universities in the area about original urban research on “Government agencies on the whole aren’t great in the cutting edge of social science and public policy. For responding to unsolicited ideas and opportunities, and us, “this is a matter of digging a little deeper,” Oates says. they are engineered to put tightly prescribed solutions out “We met with interested researchers, technologists and to bid,” Oates says. But “we have a model that allows community members to figure out how we could use data us to respond quickly to opportunity.” The city is now to improve quality of life issues. Being able to connect collaborating with the Commonwealth of Massachusetts newly available city data with some of the real big data to deliver a Citizens Connect-like mobile app that will out there in the world can make city government more engage citizens and produce important performance data. proactive—and more effective.” Boston officials are also talking with cities around the Planning for Tomorrow world about similar initiatives. “All of us in city govern- Oates contends that what has been accomplished so far ment are looking at ways to our data move valuable,” is just the beginning. By using information to interact better Oates says, “and to be part of these exciting cross-ju-

3. Bertolucci, Jeff. “Smartphones, with neighborhood leaders and citizens, for example, the risdictional initiatives.”• Big Data Help Fix Boston’s Potholes.” InformationWeek, July 25, 2012. city can respond faster and better to local events, enabling http://tinyurl.com/a9nfkdp Joe Mullich is a freelance business and technology 4. Boston Area Research Initiative. it to more efficiently deploy police to ensure public safety. http://tinyurl.com/kgu2q46 writer based in Sherman Oaks, Calif. 5. New Urban Mechanics: a city “Our ability to engage in those conversations is a critical movement focused on civic innovation. http://tinyurl.com/a3jodpv part of successfully using technology,” he says. This research project was funded by a grant from SAP.

2 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

City of Cape Town, South Africa at a glance Second largest city in South Africa, whose government was forged in 2000 by a consolidation of seven municipalities into one “unicity”

Industry: public sector

Population: 3.7 million

Size: 950 square miles

City employees: 25,500

Number of police stations: 60

Number of fire stations: 29

www.capetown.gov.za

Source: City of Cape Town

Page 25 RESEARCH UPDATE | Real-Time Enterprise Case Studies

Real-Time Data Speeds Services in “The Mother City” of Cape Town With in-memory computing and analytics tools, Cape Town is improving public safety and energy/water management at the same time. BY JOE MULLICH ape Town, the second largest city in South Africa, has long been called “The Mother City.” CAndre Stelzner, chief information officer for Cape Town, provides a tongue-in-cheek explanation for the nickname: “People say that’s because everything here takes nine months” to accomplish.

Jokes aside about the slow pace of government, Cape FIGURE 1 Cape Town at a Town realized that many pressing issues—such as Tech Hit Parade Respondents were asked: Currently, how important are each Glance improving public safety and managing water and energy use more effectively—required faster action than is of the following capabilities to your agency, institution or } Population: 3.7 million organization now and in two years? (percent of respondents typical of large bureaucracies. Research by the city’s IT } Size: 950 square miles indicating “important”) team indicated that success could only be achieved by } City employees: 25,500 collecting and analyzing real-time data. Currently Important Important in Two Years } Number of police stations: Share data and analytics across organization 60 85% 89% } Number of fire stations: Cape Town’s government, which was forged in 2000 by a 29 consolidation of seven municipalities into a single “unicity,” Provide field staff with data from internal systems on www.capetown.gov.za/ delivers all the services that are used by its population apps 84% 87% Source: City of Cape Town of 3.7 million. This includes electricity, water, healthcare,

sanitation, transportation, refuse collection, libraries, and Offer constituents access to operational systems police and fire protection, to name a few. 83% 85%

Ability to understand stakeholder sentiment via social In 2003, the city started down a path toward better media analysis use of information by implementing an enterprise 72% 81% resource planning (ERP) system. It is now one of the largest ERP deployments in any city government in the Ability to correlate location and other asset data 73% 77% world, encapsulating 450 business processes. Tangible benefits include providing insight for reducing debt and Communicate with stakeholders via social media enabling Cape Town to achieve an A+ credit rating. The 73% 74% need for faster response times has led the city to big Real-time sensor feeds from assets data analytics and in-memory computing, which can 61% 71% analyze huge amounts of data quickly. Base: 103 executives at small, midsize and large public agencies worldwide Source: Bloomberg Businessweek Research Services 2013 Agency Analytics Study It Starts with Smart Metering Cape Town’s initial foray into in-memory computing For example, “If we have real-time information about will most likely come from the need to analyze large the supply and demand for water and power, we quantities of data gathered by smart meters that can keep them in sync with dynamic tariffs,” Stelzner monitor water and energy use. Currently, the system is says. By raising tariffs in times of rising demand, the limited to large corporate users, but over the next two city can encourage conservation and reduce the risk years it will expand significantly and result in a host of of unstable power supplies that can lead to national new services. blackouts. “We need to find ways to get people to

change their behavior about the amount of water and electricity they use,” he says. Cape Town’s Challenges } Improve water/energy conservation Real-time access to water and energy data will provide } Provide better waste management to over one other benefits, as well. “We will be able to pick up a leak million households in a main water line immediately,” Stelzner says. “In the } Increase citizen safety and provide “intelligent policing” past, it could take us days to identify a leak in a remote area, which resulted in a significant amount of spillage.” Cape Town’s IT Solutions Cape Town’s } Implement real-time smart metering Government For waste management, trash containers will be } Equip trash containers with RFID tags Milestones equipped with RFID tags so the citizens can be charged } Consolidate emergency services/data into a } Fiscal management for garbage collection based on the weight of their refuse. single platform and A+ credit rating “At the moment, everyone pays the same, so the costs } Implement big data analytics and in-memory computing } Use of big data are not divided equitably and there is no incentive for analysis to conserve people to behave more efficiently,” Stelzner points out. water and energy } Use of RFID tags Consolidating Emergency Information are many of the types of activities already underway in to optimize waste Stelzner sees what may be an even bigger payoff by Cape Town, including sharing data and analytics across management applying in-memory computing to emergency services. the organization and correlating location with other data } Near-real-time At present, the city’s emergency services—from law assets (see Figure 1, “Tech Hit Parade”). response to public- enforcement to disaster management—tend to operate safety events in silos. As a result, information from one department that In Cape Town, the use of in-memory computing is being could potentially be used by another is not shared. To expanded to also enable more “intelligent policing,” solve the problem, the city is developing an emergency meaning the city can use more data sources to respond control platform that will consolidate data sets, leveraging more clinically to security events. “In some areas, security the power of in-memory computing across agencies. issues are a perception,” Stelzner says, “and in some cases, they are a reality.” If by using real-time sensors As an example, the city currently collects information and other sources “we have the ability to get a better on 1.5 million non-emergency events a year. “There understanding of the true situation, then we’ll be able to is a lot of information that is not now visible to the apply resources more effectively,” he says. emergency controller, even though this information is interrelated to the work they do,” Stelzner says. He As an example of future public safety benefits, Stelzner goes on to explain: “A traffic probe that shows highway uses a fireman and a burning building. The fire captain traffic has slowed to almost zero speed could inform back at headquarters will be able to pull up the you of another event, like a major accident before the building plan from the land-use management system, accident is reported. By consolidating this information instantly locate potentially dangerous gas cylinders in on a single platform, we can be more proactive and not the structure, note how to avoid them and then alert depend entirely on the community to report events.” the fireman on his handheld device as he enters the building. “That is the future we envision,” Stelzner says. The idea of sharing access to real-time public data and “It’s a future that depends on enormous and rapid analytics across public agencies is not unique to Cape computing power—because you can’t wait for a batch Town. In fact, some 81 percent of managers among job to run when a building is on fire.”• government entities around the world responding to Joe Mullich is a freelance business and technology a 2013 global survey by Bloomberg Businessweek writer based in Sherman Oaks, Calif. Research Services strongly agree that big data is crucial to meeting their mission. High on their list of requirements This research project was funded by a grant from SAP.

2 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Commonwealth Bank of Australia at a Glance Australia’s leading provider of integrated financial services

Industry: banking

2012 revenue: $47.2 billion

Profits: $7.09 billion

Number of employees: 51,000

Markets: Australia, New Zealand, China, Vietnam and Indonesia, with branches in New York, Tokyo and Hong Kong www.commbank.com.au

Source: Commonwealth Bank of Australia

Page 28 RESEARCH UPDATE | Real-Time Banking Case Studies

CBA Offers More Personalized Banking Through Big Data and Analytics With new data sources and tools, Commonwealth Bank of Australia is providing more personalized service to customers and building loyalty within the community. BY TOM GROENFELDT ommonwealth Bank of Australia (CBA) knows a lot about its customers. That is partly because Cit has a huge footprint in the country—it processes nine million transactions per day, handles 40 percent of the card transactions in Australia and maintains 12 million account profiles. More important than company size or number of transactions is how CBA uses its wealth of customer data, along with

real-time analytics, to build loyalty and provide superior Commonwealth customer service. Key Features of CBA’s New Web Site Bank of Australia } Magazine-style section provides content based at a Glance on customers’ key life events Like CBA, “Most banks have incredibly rich data sets, } Description: Australia’s } Integrates data from social interactions, on-site especially those derived from payments and credit data,” leading provider of behaviors and browsing history integrated financial notes Andy Lark, CBA’s outgoing chief marketing officer. } Support forum where customers seek support services Far fewer have learned how to combine this “big data” and advice about the bank’s products and services } 2012 Revenue: $47.2 with other information resources and then put it to work to } Uses principles of responsive Web design to look billion create better products and services. good on all devices, including PCs, smartphones } Profits: $7.09 billion and tablets What has put CBA out front in this area is a recent } Number of Employees: 51,000 modernization of its IT systems, followed by strategic investments in social media, analytics and mobility. All of that prospect has been looking at properties, the site is able } Markets: Australia, New to display more relevant offers—home loan, insurance and Zealand, China, Vietnam this has unified the bank’s data foundation and “put us and Indonesia, with in an enviable position,” Lark says. Thanks in part to big other related product information. The site can even analyze branches in New York, data and real-time analytics, CBA has reduced check a customer’s Web searches to provide a very specific offer Tokyo and Hong Kong fraud by 50 percent and Internet fraud by 80 percent. on the spot, in real time. www.commbank.com.au Equally important, he says, is how analytics now “gives our Source: Commonwealth Bank customer service the advantage of clarity of purpose.” The CBA site is a good example of how customers want of Australia to communicate with banks on their terms, when and All banks strive to provide superior customer service. where they are ready. According to Andrew Hagger, CBA made real progress once it realized customers general manager—head transformation and analytics at want individualized attention. “I guess one of the largest CBA, “Analytics underpins the ability to offer the right surprises to us was the extent to which customers wanted products to clients.”1 the site personalized to them,” Lark says of the bank’s Web service, NetBank. “So the strategy became making More Than a Smarter Web Site the entire site relevant and meaningful for the customers.” But CBA’s focus on customer service extends well beyond the Web. The more accounts a customer has with the Visitors can see the difference. For example, say a prospect bank, the more helpful CBA can be with pricing and has been viewing properties online and then visits the CBA advice. If a household has its mortgage, checking, savings site. Not too long ago, “you might have gotten a banner with and credit cards with CBA, the bank’s analytical tools can 1. Wisniewski, Mary. “5 Takeaways from SAP’s Client Conference: Reporter’s a standard offer for a travel money card—something far from suggest the most appropriate products and offer pricing to Notebook.” American Banker, October 25, 2012. http://goo.gl/noH0qI interesting at that point in time,” Lark says. Today, knowing fit the customer’s needs, both in person and online. CBA

can even offer them a report on their spending through a visual and interactive hub site called Signals.2 CBA’s Products and Services } Retail Banking: Home loans, credit cards, personal loans, transaction accounts, and To be customer centric, it is important to provide access demand and term deposits to data across the enterprise to individual users at the } Commercial Banking: Business loans, point of decision-making, in addition to having clean equipment and trade finance, and rural and data and an integrated analytics platform that enables agribusiness products collaboration across the various departments and } Corporate Financial Markets Services: processes. CBA’s IT infrastructure demonstrates this well. Securities underwriting, trading and distribution, corporate finance, equities, payments and “We have the strength of the relationship data and the transaction services, investment management customer-centric architecture, which I believe will set us and custody services apart from the competition on an enduring basis,” says CBA CIO Michael Harte.3 CBA has also been working on incorporating innovations such as in-memory computing to power up its strategy. that, together with analytics, runs well and looks good on desktop, tablet and smartphone devices. Harte says the bank has developed a strategy to collect transaction and other data, analyze it and use it to determine Conclusion the right pricing and financial strategies for an individual or As CBA demonstrates, the future is bright for banks that an entire household. As Lark puts it, “The more information learn to use big data and analytics to gain a competitive you have, the more inclined you will be to make better edge. However, dealing with all that data is not always informed financial decisions.” The ability to forecast trends, easy and progress can take time, says David Tanis, model options and predict outcomes is another important manager of information systems and frontline analytics at aspect of gaining customer trust and loyalty. CBA. At the Gartner Business Intelligence & Information Management Summit 2013 in Sydney in February, Tanis Increasingly, more customer and prospect data is coming said every analytics project is a learning experience, from sources outside the bank. An early and active user of especially when it involves big data.4 social media, CBA has dedicated social media teams who monitor social sites 24 hours a day looking for prospects “We’re still trying to understand what the capabilities and who might be dissatisfied with another bank and seeking skills are in terms of pulling as much out of the big data a new financial services home. The teams also look for as we can,” Tanis says. CBA deals with the wide varierty potential problems within CBA and respond to them of big data sources every day as it tries to derive business quickly. “I was surprised at how much benefit we could get intelligence from that data. from social data,” Lark says. According to Lark, many institutions are still in the early Skyrocketing use of smartphones and other mobile stages “of being able to fully utilize their data” to develop devices is also quickly filling the bank’s data coffers. CBA insights and better meet customer needs. “It’s not just the saw its mobile users jump from 19 percent to 32 percent new data sets, or the size of the data, that is driving this,” in less than a year. “Mobile is going to be even more he says. “It’s also the awareness that in the future, you’ll significant in the future, as more and more happens on need to meet the customer before they start the search. 2. Langlois, Christophe. “BIG DATA: Top Australian Banks Launch Social Media the smartphone,” Lark predicts. “Being relevant isn’t just That’s where the winners will win.” • Spending Comparison Sites.” Visible Banking, March 7, 2013. going to be about the relevant offer; it also means being http://goo.gl/0MEoxE Tom Groenfeldt is a freelance writer in based in 3. “We’re changing the game: contextually relevant to where I am and what I am doing.” CommBank.” The Sydney Morning Herald, Sturgeon Bay, Wis., who has written for many financial, May 29, 2012. http://goo.gl/zEq8w 4. Lui, Spandas. “Half of analytics business and technology publications. investments will be a waste: CBA uses a combination of ad serving, smart media Commonwealth Bank.” ZDNet, February 25, 2013. http://goo.gl/h385jv buying and personalization engines on its mobile Web site This research project was funded by a grant from SAP.

2 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

ConAgra at a glance Consumer products giant produces brand-name foods such as Healthy Choice, Marie Callender’s and Orville Redenbacher’s, among others

Industry: consumer products

Category leaders: 27 of its consumer brands are No. 1 or No. 2 in their category; 23 consumer brands generate more than $100 million in retail sales each year

Founded: 1919

Headquarters: Omaha, Nebraska

Sales: $18 billion

Employees: 36,000 www.conagra.com

Source: ConAgra

Page 31 RESEARCH UPDATE | Real-Time Business Case Studies

Real-Time Business Leads ConAgra to Profitable Insights The consumer products giant is reducing costs, improving pricing and forging closer relationships with retailers using in-memory computing.

BY JOE MULLICH onAgra, the $18 billion consumer products (CP) giant, produces brand-name ConAgra at a Cfoods such as Healthy Choice, Marie Callender’s and Orville Redenbacher’s, Glance among others, that can be found in almost everyone’s refrigerator or pantry. Yet it Segments: Consumer foods, private brands, commercial foods faces the same challenges that pressure the entire FIGURE 1 How Analytics Contributes to Category leaders: 27 of food industry: figuring out the optimal pricing for its Business Decisions its consumer brands are products in an environment where consumers are Areas where food and beverage companies use data and No. 1 or No. 2 in their hyper-sensitive, while coping with the ever- analytics to support strategic decisions (percent responding) category; 23 consumer fluctuating costs for 4,000 raw materials used in Customer insights brands generate more some 20,000 products. 59% than $100 million in retail sales each year Brand and product management 1 Founded: 1919 According to a 2013 study by KPMG, 51 percent of 58% Headquarters: food industry executives said pricing pressures Omaha, Neb. remain the highest barrier to growth, up from 42 Pricing decisions Sales: $18 billion 56% percent a year earlier. Volatile commodity prices— Employees: 36,000 the amount manufacturers pay for ingredients—are Operating model optimization www.conagra.com another major obstacle. 52% Source: ConAgra Market expansion “If the price of beef goes up on the commodity 43% exchange, what is that going to do to our margin?” asked Mindy Simon, ConAgra’s vice president of IT. Portfolio rationalization 42% Being able to answer such questions is of

paramount interest to the company, because Source: KPMG 2013 Food and Beverage Industry Outlook Survey. http://goo.gl/me3va8 commodity prices affect its earnings and its stock price. In the past, ConAgra has warned that ConAgra decision-makers have real-time insight into earnings would be impacted by materials prices that the company’s costs and consumers’ demands. were rising faster than expected.2 “We’ve developed much better capabilities in pricing So ConAgra puts great emphasis on optimizing its analytics as part of an overall revenue growth prices, which in turn requires getting ever better management approach,” ConAgra CEO Gary

1. KPMG. 2013 Food and 3 Beverage Industry Outlook Survey. prices for commodities such as wheat, corn, oats, Rodkin told analysts in February 2014. Rodkin’s PDF. http://goo.gl/me3va8. soybean meal, soybean oil, meat, dairy products and remarks echoed the KPMG survey findings: Food 2. Rappeport, Alan. “ConAgra hit by surging raw material costs.” sugar. The company turned to in-memory computing, and beverage executives reported they rely more on Financial Times. Sept. 20, 2011. http://goo.gl/yih35b. which loads extremely large volumes of data from data and analytics to support their pricing decisions, 3. Thompson Reuters Streetevents. “ConAgra Foods at CAGNY multiple sources into one database and enables understand customers better, and support brand Conference. ” Edited transcript. users to answer questions almost instantly. Access to and product management decisions (see Figure 1, Feb. 18, 2014. http://goo.gl/QGrc9a. great volumes of rapidly processed data means “How Analytics Contributes to Business Decisions”).

ConAgra’s Real-Time Data Means Faster Decisions FIGURE 2 Consumer Products Companies’ Challenges ConAgra has priced products appropriately when it Data Challenges } Gain in-the-moment delivers value to consumers and earns a profit. For They want timelier information. (percent responding) insight into the cost of example, the company has 140 meal products, such as raw materials. Timeliness for decision-making potpies, that sell for less than $3. “Banquet potpies sell 69% } Acquire more accurate consumer purchase for less than $1, and Marie Callender’s potpies are Quality around $2.50,” Rodkin explained.4 Having a variety of information to improve 67% merchandizing. products at different price points makes ConAgra } Forge closer ties with competitive within the categories it serves, as does Integration retailers. doing a better job of getting retail prices right than its 57% competitors. Any price increases prompted by rising ConAgra’s Availability Solutions manufacturing costs have to take into account 53% } Implemented in-memory consumer expectations, observed Bob Nolan, technology to speed Source: Accenture, “Building an Analytics Driven Organization,” 2013. ConAgra’s vice president of customer insights and http://goo.gl/KBfJnv analysis. analytics. Consumers might be willing to pay 10 percent } Gathered new sources more for a gallon of gas, he said, but not for popcorn.5 Improving their ability to model alternative business of data about consumer behavior. scenarios is becoming a critical goal for CP } Shared data-driven CP companies have traditionally approached key companies that want to stay ahead. A recent insights with retailers. business decisions, such as pricing, through a Aberdeen Group study10 looked at how behavior periodic planning process. However, as the cost of differs among companies that are aware of the manufacturing and consumer demand become more business drivers, such as costs or pricing, that volatile, so do revenues. Business risks also impact their performance compared to those who increase. Companies now realize that planning are not aware. The results found companies that should be a continuous activity based on real-time take business drivers into account when creating information. Yet the majority of CP companies are their business plans are nearly three times as likely failing, thus far, to put real-time analytics at the heart to make what-if analyses a priority. of their decision-making processes, according to a new Accenture study.6 The top challenge for CP By focusing on a key pain point—costs—ConAgra companies, cited by 69 percent, is lack of timely has convincingly demonstrated the business value of 4. Thompson Reuters data (see Figure 2, “Consumer Products Companies’ its in-memory computing investment, observed Chris Streetevents. 5. Nolan, Bob. “ConAgra’s VP of Data Challenges”). Cicarello, senior director, pricing and customer Analytics on Big Data & Careers in analytics at ConAgra.11 Choosing an important Retail.” Video. Jan. 9, 2014. http://goo.gl/PS1ozI. To gather more timely data and enable faster business problem to address, as ConAgra did, is a 6. Accenture. “Building an Analytics-driven Organization: decision-making, ConAgra turned to in-memory savvy strategy for CP companies that are engaging in Organizing, Governing, Sourcing and Growing Analytics Capabilities computing in numerous areas of its business. The cutting-edge analytics. According to Accenture, in CPG.” PDF. Feb. 19, 2014. companies often launch such efforts without knowing http://goo.gl/KBfJnv. “most revolutionary” use now is forecasting materials 7. Henschen, Doug. “Can SAP costs, Simon offered.7 what they want to accomplish, and they end up Master Cloud & On-Premises?” InformationWeek. Feb. 5, 2014. spending a lot of time—and money on technology— http://goo.gl/G1yJoA. 12 8. SAP. “SAP HANA: An In- ConAgra has been able to speed up the collection of without having much impact on the business. Memory Data Platform for Real- Time Business.” PDF. 2012. its data related to commodities purchases from 9 hours http://goo.gl/gDicpN. to 20 minutes8 and to reduce its month-end forecasting Capitalizing on New Data Sources 9. Doug Henschen. 10. Castellina, Nick. “Don’t Fly process by three days. By collecting and analyzing the Another reason why ConAgra needs to process Blind When Budgeting and Forecasting: The Value of Being data more quickly, the company is able to consider massive amounts of data quickly: the company Driver-Conscious.” Aberdeen PDF. uses more types of data in its analysis. For November 2014. timelier what-if scenarios, giving it more ways to adjust http://goo.gl/7VWSGj. its strategies for buying materials, observed Simon9 example, the CP giant is adding shopper-specific 11. Bob Nolan. 12. Accenture. (see Figure 3, “Better Decisions Through Data”). data that it gets directly from retailers and

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

analyzing that data in new ways to better FIGURE 3 Better Decisions Through Data understand consumers and respond to their Business drivers for investment in analytics needs. That goes beyond how ConAgra uses include the following. (% responding) internal data to forecast demand and increase Improving the quality of decision-making sales, and speaks to how it applies aggregated 59% retail data to plan promotions and strengthen its Making quicker decisions The Benefits of supply chains. 53% Real-Time Business at As ConAgra searches these new data stores, it is Improving planning and forecasting 47% ConAgra finding surprising insights that enable better } Expanded options: product merchandizing. For example, the ConAgra is able to Developing new products, services and company learned that people who buy one type revenue streams analyze multiple “what-if” 47% scenarios when of single-serving frozen food tend to buy several

considering its strategies other single-serving varieties at the same time. Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX for purchasing Because of that, ConAgra suggested that retailers commodities. group the packages with smaller portions Enhancing Relationships } Better merchandizing: together, instead of stocking single-serve pizzas To develop a strong relationship with retailers, Using external data, with family-size pizzas. ConAgra realized that it could not simply provide such as consumer data provided by retailers, self-serving data about its own products. Nolan delivers real-time insight Such findings enable ConAgra to forge stronger pointed out that data sharing requires into how to merchandize ties with retailers: The company is giving the transparency: Retailers should fully understand products more effectively retailers data that, in turn, enables them to improve why ConAgra is making recommendations about and increase margins. their own operations. The deeper understanding which products to put on the shelf. In some cases, } Closer relationships about prices ConAgra gains not only helps the the company’s recommendations might even with retailers: Sharing data-driven company manage itself efficiently, it also helps include adding a competitor’s products. “We try to recommendations with ConAgra guide retailers on how to price products be objective and give recommendations that allow retailers helps ConAgra on the shelf. That, along with controlling costs, is [retailers] to grow their entire store, not just our build trust and become the other crucial element in maximizing profits, business,” he said. “It’s almost like being a an indispensible partner. according to ConAgra executives.13 consultant. We want to become the indispensable partner to our customers.”14 Aberdeen Group Analyst Nick Castellina said best-in-class companies—the ones that shape Nolan said this objectivity demonstrates to their plans around business drivers—are twice as supermarkets and retailers that ConAgra’s data- likely to share information with suppliers, driven suggestions are based on what will most customers, resellers and regulatory bodies. But the benefit the retailers’ bottom lines. This tactic builds data has to be accurate and recent, or it is not trust as ConAgra develops relationships with the useful. The high-performing companies that retailers. As a result, he thinks ConAgra will profit and Aberdeen identified as sharing information are “win the ties” when retailers debate whether to add nearly three times more likely to have real-time ConAgra products or those from another company.15 financial metrics. It is another innovative way that data drives “Having accurate demand forecasting and ConAgra’s sales and boosts the bottom line. • inventory is important in consumer goods,” Joe Mullich is a freelance business and technology Castellina said. “This is especially important in writer based in Sherman Oakes, Calif. 13. Bob Nolan. the food industry because of shelf life and the 14. Bob Nolan. 15. Bob Nolan. capability for materials to go bad.” This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

eBay at a glance Leading online retailer provides consumer-to- consumer sales services via the Internet

Industry: high tech Founded: 1995

Headquarters: San Jose, California

Brands: eBay, PayPal, eBay Enterprises, StubHub

Buyers worldwide: 145 million

Active users worldwide: 128 million

Number of items listed for sale: 650+ million 2013 revenue: $16.0 billion www.ebay.com

Source: eBay

Page 35 RESEARCH UPDATE | Real-Time Business Case Studies

Big Benefits from Big Data at eBay Online marketplace provider eBay uses predictive analytics and in-memory computing to better serve customers and bolster the bottom line. BY LARRY LANGE

nline retailers face two big challenges: improving customer service, and keeping Opartners and suppliers satisfied. Both are vital, because online competition is fierce. If eBay at a Glance shoppers aren’t happy with their site experience—from browsing to buying—they can easily } Founded: 1995 } Headquarters: San Jose, go elsewhere; competitors are just a click away. It’s FIGURE 1 Predictive Analytics: Calif. the same for an online retailer’s suppliers. If they } Brands: eBay, PayPal, What Is It Good For? don’t like certain terms and conditions, or consider eBay Enterprises, (percentage of survey respondents) technical features too basic, they won’t stay loyal. StubHub Achieve competitive advantage } Buyers worldwide: 145 68% million The stakes for online retailers are huge. Worldwide } Active users worldwide: B2C ecommerce sales will increase more than 20 New revenue opportunities 128 million 55% } Number of items listed for percent this year, reaching $1.5 trillion, predicts sale: 650+ million research firm eMarketer.1 Increased profitability } Revenue (fiscal 2013): 52% $16.0 billion Leading online retailer eBay is meeting both www.ebay.com challenges by deploying business analytics to provide Increased customer service Source: eBay online sellers with real-time information about their 45% eBay-enabled sales. Sellers can use this information Operational efficiencies to sell more goods more efficiently and effectively. 44% Because eBay is better serving its sellers, it is able to indirectly better satisfy buyers with a speedier, Source: Ventana Research predictive analytics benchmark research, 2012. http://goo.gl/k5lkGk optimized shopping experience. As for eBay’s bottom line? It gets a cut of every sale, regardless of which and services that empower customers to perform a variety customer is buying or selling. Simply put, the new of business tasks (see Figure 1, “Predictive Analytics: What analytics program is a win-win-win. Is It Good For?”).

Analyzing Analytics Many organizations use predictive analytics (see Figure The move is significant for another reason: 2, “Predictive Analytics: What It Is Used For”). Netflix Although many ecommerce companies find the commissioned two TV shows, “Orange Is the New promise of big data compelling, few have realized Black” and “House of Cards,” knowing beforehand, its business benefits. Now, though, early adopters thanks to predictive analytics, that a big chunk of its like eBay are achieving new revenue streams, customer base would loyally watch them.2 greater speed-to-market and enhanced business flexibility. Among the approaches yielding results, A mining company uses predictive analytics to better predictive analytics stands out. maintain equipment. The company, as described by Henry

1. eMarketer. “Global B2C Ecommerce Morris, senior vice president of software and services Sales to Hit $1.5 Trillion This Year Driven by Growth in Emerging Markets.” Feb. 3. Predictive analytics helps companies extract research at market researcher IDC, deploys Caterpillar 2014. http://goo.gl/K8l0Vt. 2. Carr, David. “Giving Viewers What information from mountains of data, forecast customer machinery that, in the past, was manually checked and They Want.” The New York Times, Feb. 24, 2013. http://goo.gl/Tr4wAw. trends and patterns of behavior, and create products maintained only on an infrequent and irregular basis.

Now equipment is continuously monitored. “The FIGURE 2 Predictive Analytics: company can detect a ‘signal,’ or datapoint anomaly, on What It Is Used For a machine that predicts when the machine is likely to fail,” (percentage of survey respondents) Morris explains. “Then they can send someone out to Currently using Plan to add in future Forecasting repair it.” 72% 24% Speed is an important benefit of predictive analytics. A Benefits of new, faster technology, known as in-memory computing, Marketing 67% In-Memory keeps data in a computer system’s RAM rather than 22% Computing at eBay storing it on traditional physical disks (see Figure 3, } Sellers get relevant, “In-Memory vs. Disk: The Shootout”). In-memory Customer service real-time information computing helps predictive analytics systems aggregate 45% about their own 34% customers and eBay and analyze vast amounts of data from multiple data buyers sources in nearly real time. Anand Rao, a principal at Product offers } More in-depth data consultancy PricewaterhouseCoopers (PwC), says 43% analysis; faster in-memory computing can deliver “increased innovation, 22% anomaly detection and notification improved productivity, enhanced customer experience Fraud detection } Dashboard lets and loyalty, and lower costs.”3 34% analysts visualize 31% relevant information in Many enterprise software providers offer automated several categories Source: Ventana Research predictive analytics benchmark research, 2013. platforms based on in-memory computing. These http://goo.gl/1kUhOz } Tasks that previously took a month can now systems run data through myriad preconfigured be done in one day metrics, quickly send those metrics to users and To stay competitive, eBay employs 5,000 data analysts } Data is delivered provide dashboards for users to disseminate data. worldwide who sift through as much as 100 petabytes of to the entire team Researchers at audit and consulting firm Deloitte say data. Until recently, analysts used Microsoft Excel of eBay analysts new in-memory platforms offer query response times spreadsheets. But the process wasn’t scalable, and instantaneously that are thousands of times faster than more errors were common.6 Given the huge amount of data conventional approaches; transactions can be and eBay’s pressing business needs, the case for processed as much as 20,000 times faster.4 automation was easily made.

These and other related benefits are making predictive eBay’s first test project, launched in 2013, combined analytics a serious business. Research firm predictive analytics and in-memory computing. The MarketsandMarkets forecasts the global market for result empowered the North American marketplace predictive analytics systems reaching $5.24 billion in team of 300 data analysts to provide sellers with 2018, up from $1.7 billion last year.5 forecasts, as well as relevant and real-time information about sellers’ own customers or eBay buyers. Up, Up and eBay Information was provided as actionable metrics, 3. PricewaterhouseCoopers. “6th Annual Digital IQ Survey.” http://goo.gl/1dQiyn. eBay recognizes the value of predictive analytics and including customer tastes and preferences, market Comments by Anand Rao in “Digital IQ 2014 10 Technology Trends for Business.” in-memory computing. It uses the technology to score conditions, and timely supply and demand information. http://goo.gl/IpJsLw. 4. Deloitte University Press. “Tech Trends both business and technology benefits. 2014: Inspiring Disruption.” http://goo.gl/E3f6z3. Consider the popular category of collectible men’s 5. MarketsandMarkets. “Predictive Analytics Worldwide Forecasts Data at eBay is certainly big: It serves some 20 million basketball sneakers. “If we know that ‘collectible (2013–2018).” http://goo.gl/4fEhBg. 6. SAP. “Optimizing eBay’s Signal sellers offering wares in more than 4,500 product sneakers for the Olympics’ are selling really well, we Detection with SAP HANA.” http://goo.gl/UFHy7j. categories and 145 million active buyers in 100+ want to tell our sellers as soon as we know it,” says 7. Schwarzbach, David. “SAP Sapphire Conference Presentation: countries. Now eBay is feeling pressure to keep David Schwarzbach, vice president and CFO of eBay Target Business-Critical Processes for 7 Transformation.” http://goo.gl/uIv6Zu. business competitive and customers loyal. North America. “That’s the value of speed.” The

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

sellers, equipped with this FIGURE 3 In-Memory vs. Disk: The Shootout knowledge, can then run special How these two technologies compare on two important factors. promotions for those sneakers. Infrastructure Finding Data In-memory Data is stored in the Streamlined data retrieval Keys to the Kingdom computing machine’s main system and processing means no The eBay team had worked alone on a memory, enabling mechanical parts, no scalable and configurable automated in-line processing and movement and no need eBay’s Business predictive analytics system for two years. manipulation. for alignment. Challenges But the payoff came when } Data analysis was Schwarzbach’s team began working Disk-based Data is mechanically For data to be read or recorded on a spinning written, the correct manual; the process with an in-memory platform vendor. wasn’t scalable, and disk, often on a large address must first be “There was a thrilling, magic moment,” errors were common array of networked found on the physical } Difficult to coordinate he says, when the team knew they storage appliances. disk. Overhead for input/ analyst’s efforts, could transfer their data analysis output activities can diagnose problems operation over to an in-memory platform. cause delays to jobs that and make accurate Schwarzbach adds: “We’d cracked it.” are large or complicated. decisions on seller recommendations Source: Deloitte, “Tech Trends 2014: Inspiring Disruption” By deploying the in-memory platform, eBay’s Solutions eBay honed in on two core requirements: greater a method to deliver items before a customer orders } Deploy advanced analysis of data, and faster detection and notification of them. In deciding what to ship, Amazon will leverage its business analytics anomalies—or signal detection. The platform delivered trove of customer data—and its predictive analytics } Combine predictive system—to take into account a customer’s previous analytics and in- several tools, including a dashboard that lets eBay memory computing North America’s analysts visualize relevant information in orders, product searches, wish lists and more.10 } Partner with in-memory several categories and that flashes data anomaly alerts. platform vendor Similarly, UPS wants to deploy prescriptive analytics on Thanks to in-memory computing, tasks that previously all its data feeds, including GPS tracking, driver routes took a month now take a day. Gagandeep Bawa, a and workloads, traffic patterns and delays, and manager in eBay North America’s financial planning customer delivery requirements. The goal: a real-time, group, says this real-time speed frees up data constantly adapting operational system that optimizes analysts’ time, and he calls the system “inherently any task performed by the company’s drivers.11 intelligent and configurable.”8 PwC says the future of enterprise applications will Next, eBay will offer this data directly to sellers, either combine predictive analytics with mindflows—basically, on their personalized Web pages or via a mobile app. a person’s thoughts—to create “mindful apps.” These Schwarzbach says his team is “taking this to apps will combine data generated by tracking what production right now.”9 Morris of IDC expects this will people think, their behavior patterns and data already 8. Bawa, Gagandeep. “eBay’s Early Signal Detection System Runs Machine give eBay a big competitive edge. “Other auction firms being aggregated by a firm’s predictive analytics Learning & Predictive Powered by SAP HANA.” http://goo.gl/282uJB. don’t offer this kind of data analysis as a service,” he program. That information will help employees make 9. David Schwarzbach. 12 10. Bensinger, Greg. “Amazon Wants to explains. “So this makes eBay very compelling for optimal decisions, all served up via a mobile app. Ship Your Package Before You Buy It.” The Wall Street Journal, Jan. 17, 2014. customers to use for selling their goods.” http://goo.gl/Vz1U9i. 11. Hemsoth, Nicole. “UPS Delivers on If you were using one of these programs, you’d already Prescriptive Analytics.” Datanami, Feb. 23, 2013. http://goo.gl/PP4GjV. Schultz, The Future Is Predictive know that! • Beth. “Inside Analytics: UPS Delivers the Goods.” All Analytics, May 9, 2013. Impressive as these gains are, several companies are http://goo.gl/iEfm4U. Larry Lange is a freelance business and technology 12. Baya, Vinod, Galen Gruman and Bo pushing new developments that will make data Parker. “The future of enterprise apps: writer based in Jacksonville, Fla. Moving beyond workflows to mindflows.” analytics even more useful. Amazon was recently PricewaterhouseCoopers. http://goo.gl/GqCOLH. granted a patent for what it calls anticipatory shipping, This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

EMC at a glance IT as a service, cloud computing. 25 distinct product families span hardware, software and services, addressing storage, trusted infrastructure, cloud computing and big data

Industry: high tech

Headquarters: Hopkinton, Massachusetts

Founded: 1979

2013 revenues: $23.2 billion

Acquisitions to date: 70

Employees: 60,000 in 86 countries www.emc.com

Source: EMC

Page 39 RESEARCH UPDATE | Real-Time Business Case Studies

The Key to EMC’s Technology Growth: Integrated Acquisitions Integrating the systems and business processes of acquired companies is often overlooked. EMC, with its 70 acquisitions and counting, has a winning formula.

BY ALAN RADDING he rise of so-called SMAC technologies—social, mobile, analytics and cloud— T presents a new challenge to large technology suppliers. On the one hand, these companies must respond to their customers’ rising demands for these new technologies. EMC at a Glance Yet on the other hand, doing so is costly, time-consuming and complicated. Industry: IT as a service, cloud computing To meet this challenge, many large technology FIGURE 1 Acquisitions Do Not Company description: 25 companies are turning to mergers and acquisitions Always Deliver distinct product families (M&A). They are buying smaller suppliers that have (% reporting deal objective was “very important” and span hardware, software “completely achieved”) developed solutions in one of these areas, and and services, addressing Very important Completely achieved then integrating them into their product mix and storage, trusted All respondents All respondents infrastructure, cloud operations. Among the firms that have spent Very important Completely achieved computing and big data millions, even billions, of dollars to acquire SMAC Technology respondents Technology respondents Headquarters: suppliers are IBM, Google, Microsoft and EMC. Hopkinton, Mass. Access to new brands, technologies or products Founded: 1979 56% Rob Fisher, U.S. technology industry deals leader at Revenues: 82% PricewaterhouseCoopers (PwC), an audit, tax and $23.2 billion (2013) 40% Acquisitions to date: 70 consulting firm, said these companies essentially 47% Employees: have no choice. SMAC, he said, changes “the 60,000 in 86 countries nature of the business, especially when software is Access to new markets www.emc.com involved.” Acquisitions are also a way for large 70% Source: EMC technology suppliers to enter new markets at lower 75% risk. “Upstarts are prepared to lose hundreds of 58% millions of dollars of venture capital before becoming 38% profitable,” Fisher explained. “But established Access to management or technical talent players cannot absorb those kinds of losses.” 43% 59% PwC recently completed a three-year study of 36% mergers and acquisitions in technology, 31% information, communications and entertainment. As part of the study, the firm asked C-level Source: PricewaterhouseCoopers, 2014 original rendering executives to name their most important deal objectives. Among executives in the technology percent of the technology executives (see Figure 1, sector, two goals came out on top: access to new “Acquisitions Do Not Always Deliver”). brands, technology and products, cited by more than 80 percent of respondents; and access to However, simply acquiring a smaller technology new markets, also cited by 75 percent. The next company is not a panacea. The acquired company closest was access to talent, cited by nearly 60 must also be integrated into the larger company, and

that includes its IT architecture, systems and FIGURE 2 Transformation via Acquisition applications. This work is far from simple. PwC’s (% of acquisition type, based on the largest survey identified IT integration as the number one acquisition in the past three years) post-M&A closing difficulty. It was followed by the All respondents Technology respondents related challenge of integrating operating Transformational procedures and business processes. 44% EMC’s Challenges 53% } Support both current This is essentially an information-platform challenge. and anticipated future The acquiring company must meld disparate Absorption businesses. 29% processes and technologies, most originally } Manage a large number 29% intended to support different business models, (70+) of acquisitions, Source: PricewaterhouseCoopers, 2014 and integrate business customer bases, development cycles and product- and operational systems. release expectations. And with cloud technology } Enable current products enabling a software supplier to shift its business Originally founded as a provider of data-storage with mobility and cloud model from licensed products to hosted services, systems to large enterprises, EMC is today a $23.2 computing. even the most basic accounting systems may billion (revenue) supplier of 25 distinct product families } Update legacy ERP require major adjustments. that span hardware, software and services. These system to support variable pricing, products address not only storage, but also what the especially pay-as-you-go As a result of this complexity and difficulty, many company calls “trusted infrastructure,” cloud computing models. acquisitions fall short of achieving their goals. and big data. PwC’s survey found that fewer than one-third of executives at acquiring technologies companies EMC’s customer base has expanded to now include both reported competence in integration. The results small and midsize companies. As a result, it engages both were striking: Fewer than half (47 percent) of the direct and in-direct sales channels. But its typical order executives said their acquisitions actually size has shrunk. In response, EMC embarked on a leaner delivered access to new brands, technologies and operating strategy, methodically cut costs and adopted a products—goals the deals were created to deliver. high-volume/low-touch business model, much like the As for gaining access to new markets and new best SMAC players. And at the manufacturing end, EMC talent, the results were even worse; only 38 operations now rely on more than 50 suppliers worldwide, percent said their acquisition met expectations for up from a small handful in 2004, as well as several new markets, while a mere 31 percent said it met contract manufacturers. expectations for talent. “Transformational deals are much harder due to the increased complexity Internally, EMC is a very different company, too. As of of the integration,” said Gregg Nahass, U.S. M&A 2013, EMC had 60,000 internal users, more than twice integration leader at PwC (see Figure 2, the number it had in 2004. The company now has more “Transformation via Acquisition”). than 400,000 customers and partners, six times the number it had when a previous ERP system was The Integration Hurdle deployed. Similarly, EMC’s roster of applications and tools Integrating systems, therefore, is a key differentiator has grown to 500, up from 400 in 2004. Even more between acquisitions of SMAC companies that important, the way the company handles its workload has succeed at meeting their goals and those that fail. changed dramatically; EMC now operates more than more One company aware of this challenge, and working than 10,000 virtualized systems, amounting to 93 percent hard to master it, is EMC. Over the past decade, of all its servers (see Figure 3, “EMC’s Evolving IT”). EMC has acquired no fewer than 70 companies. What is more, these acquisitions have enabled EMC With all these changes, EMC found it also needed to to radically change the nature of its business. change its internal IT environment. The ERP system

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

that had suited the company so well in 2004 was FIGURE 3 EMC’s Evolving IT by 2013 no longer serving the business. EMC now needed to accommodate dozens of business 2004 2013 units supporting varied product families being Internal users 24,000 60,000 sold via multiple channels to different types of customers. A new system was clearly needed, IT environment 5 datacenters, 5 datacenters, 960 terabytes ~13 petabytes one that could support myriad types of pricing, EMC’s Solutions of storage of storage } Move to an open ERP from conventional on-premises licenses to various Business ~400 ~500 system. subscription and pay-as-you-go models—and applications } Create six new whatever might come next. And the company principles for any future Virtualization 2,000 physical 93%, ~10,000 would need to help its staff work seamlessly. No servers OS images system. matter which EMC business units they were in, } Launch Project Propel, Source: EMC Corp. an enterprise system workers needed to be able to share information, initiative designed to collaborate and innovate with one another. transform 125+ The EMC team began with a question inspired by one of business processes. Beyond Integration its acquisitions: What if the same principles that } Deploy a set of Clearly, EMC needed a solution that went far transformed a single acquisition, delivering virtualized enterprise beyond merely upgrading its existing ERP system unprecedented value for customers, were applied to the apps that can be adapted to changing with a few cloud or mobile capabilities. Instead, the entire company? future needs. company needed a system that would integrate its } Plan to add new acquired companies and support its new business In answering that question, EMC officials realized they capabilities, including would need to abandon their legacy ERP system. The CRM, scheduled for “Now that we’re in the hardware replacement, they realized, would need to be based on 2015 deployment. business, the software business, open systems. It would need to enable widespread virtualization. And it would need to support EMC’s and very much into mergers journey to cloud and mobile computing. Above all, EMC and acquisitions, we needed needed a system that could integrate everything and to reinvent our ERP system to share information, even with future acquisitions.

support not just the business To find the right solution, EMC highlighted six key that we had evolved into but the principles: • Design for the future state of the business. businesses that we will continue • Minimize customizations and commit to the vendor’s to evolve into going forward.” upgrade path. • Move fast to mitigate system risk. —MICHAEL HARDING, PROJECT LEAD TECHNICAL ARCHITECT, EMC • Use a phased approach to realize value as quickly as possible. processes, customer types and business models. A • Recognize and manage the cultural change and conventional upgrade would not suffice. “Now that transformation the new system entailed. we’re in the hardware business, the software • Prepare the staff with new skills and competencies. business, and very much into mergers and acquisitions, we needed to reinvent our ERP system After a careful search, EMC settled on a solution supplied to support not just the business that we had evolved by an industry-leading enterprise systems vendor. The into but the businesses that we will continue to solution would support from the outset open systems, 1. “HANA Customer Spotlight.” evolve into going forward,” said Michael Harding, virtualization and cloud computing. To do that, while SAP, 2013. http://goo.gl/UtnxDU. project lead technical architect at EMC.1 ensuring performance and flexibility, the new vendor also

3 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Project Propel at brought a slew of the latest technology advances, acquired vFabric, GemFire, Cetas and Cloud a Glance including in-memory processing. Foundry). The combined service now provides data } Overall mission: warehouse protection, data deduplication and Support both EMC’s Transformation Project backup services. current and anticipated The new enterprise system initiative, referred to future businesses. internally as Project Propel, was designed to The new integration and innovation capabilities of } Key dates: Phase 1 launched in 2013; transform more than 125 EMC business processes Project Propel have also helped EMC provide Phase 2 (adding RSA by deploying 65 integrated, multi-module enterprise real-time financial reporting. “The end-of-quarter and services groups) applications. Project Propel has also enabled EMC to process [is] a huge amount of pressure,” Tony is scheduled for 2014; deploy an integrated set of virtualized enterprise Pagliarulo, EMC’s senior VP and chief operating Phase 3 (adding CRM applications that can be adapted to support both the capabilities) is scheduled way EMC works now and the way it is likely to work “The key to innovation is to for 2015. } Business benefits in the future as it pursues ongoing innovation. leverage technology for multiple include: Reduced departments. It’s being informed capital and operating Project Propel’s initial rollout in 2013 involved some 20 expenses, improved modules. These included core ERP, business that creates the basis for business performance, intelligence (BI), supplier relationship management the ability to leverage sustainable innovation.” (SRM) and advanced planning and optimizing (APO), commodity IT hardware, as well as some portals and various third-party —GREG SCHULZ, SENIOR ANALYST AND improved security and FOUNDER, STORAGEIO easy information sharing applications. Despite the project’s massive scale, it to facilitate innovation. was deployed to EMC’s internal users in just four days. officer, said at a recent investment analyst briefing. } Business applications: “So it was critical for us to have real-time reporting approximately 500. The effort was intended from the start as a at a very detailed level.” Thanks to Propel, he } Transformation: More multi-stage rollout. The last stage, expected to go added, “We were able to make that available to than 125 EMC business processes have been live in 2015, will add customer relationship users so they can make real-time decisions about added; 65 legacy management (CRM) capabilities to bring the sales which customers got what product. It was huge—it applications have been groups under the same information-sharing really saved our bacon.”2 retired. umbrella, enabling all groups to see the same } In-memory processing: information. At that point, information should flow Project Propel is the largest IT transformation in Four third-party efficiently and seamlessly through major swaths of EMC’s history. It delivers new technology, enables appliances. } Product families: the sprawling company, with innovation following EMC to redesign key business processes and More than 24, including right along. ensures an agile approach to deployments. This, in hardware, software and turn, helps EMC respond quickly and completely to business services. This information flow will facilitate another EMC customer requests, provide a better mix-and- } Training: 103 classes objective, sustainable innovation, said Greg Schulz, match of products and services, and share have been completed, senior analyst and founder of StorageIO, an IT information internally. training 1,200 “super users.” consulting firm. “The key to innovation is to leverage } Phase 1 technology for multiple departments,” he added. These types of capabilities are important, not only to implementation: 40,000 “It’s being informed that creates the basis for EMC but to the entire IT industry. As the market users at 316 sites in just sustainable innovation.” leaders continue to grow and gain capabilities with four days. acquisitions, the need to create open, integrated Source: EMC Corp. An early example of that type of innovation is EMC’s systems will only grow.• creation of Pivotal, its agile infrastructure service. Alan Radding is a freelance business and technology 2. “Transform Business EMC formed Pivotal by combining capabilities it with an Innovation-Driven writer based in Newton, Mass. Enterprise.” SAP Webcast, gained through the acquisitions of Pivotal Labs, 2013. http://goo.gl/trhTSI. Greenplum and VMware (which, in turn, had This research project was funded by a grant from SAP.

4 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Florida Crystals at a glance A subsidiary of Fanjul Corp., Florida Crystals is the world’s largest sugar refiner. A vertically integrated company, it operates two sugar mills, a sugar refinery, a packaging and distribution center, a rice mill and a renewable energy power plant

Industry: consumer products and agriculture

Headquarters: West Palm Beach, Florida Farmland: 155,000 acres

Products: organic and natural sugar; refined sugar; organic and premium rice

Raw sugar production: 800,000 tons per year

Refined sugar production: 500,000 tons per year

Annual revenues: $5.5 billion (Fanjul Corp.) Employees: 2,000 www.floridacrystals.com

Source: Florida Crystals

Page 44 RESEARCH UPDATE | Real-Time Business Case Studies

Real-Time Business Sweetens Performance at Florida Crystals In-memory computing leads to more effective decision-making and energizes an aggressive growth strategy at the world’s largest sugar refiner.

BY STEPHANIE OVERBY arly in 2014, due to global oversupply, sugar prices sank to less than 15 cents a Florida Crystals At a Glance E pound on the IntercontinentalExchange—the lowest price in three-and-a-half Industry: Agriculture and years.1 But unlike manufacturers in other industries, Florida Crystals cannot cut consumer products Company Description: production or hold back supply to compensate A subsidiary of Fanjul FIGURE 1 Real Time vs. Reality when prices fall. Sugar cane, once planted, Many companies struggle to deliver Corp., Florida Crystals in-the-moment business data in real time. (% of global is the world’s largest produces a crop for several years, and mills must companies with one-day or near-real-time visibility) sugar refiner. A vertically extract the raw sugar from the freshly harvested integrated company, cane soon after each new crop ripens. Cash positions/liquidity it operates two sugar 57% mills, a sugar refinery, Meanwhile, in a commodity industry such as sugar a packaging and Customer information/business volume that is dominated by a few large players, it is difficult 43% distribution center, a rice for firms to differentiate their products, noted Simon mill and a renewable- Working capital energy power plant. Ellis, practice director of supply chain strategies at 39% Headquarters: West Palm IDC. As a result, profit margins for sugar companies Beach, Fla. are tight and cost efficiencies are key to profitability. Supplier base/spend volume Farmland: 155,000 acres 28% Florida Crystals faces another challenge, too: Products: Organic and business complexity resulting from an aggressive natural sugar; refined Financial performance 24% sugar; organic and growth-by-acquisition strategy and vertically integrated operations that cover every step of sugar premium rice Operational risk Raw sugar production: production, from planting to distribution. 17% 800,000 tons per year Refined sugar production: Source: The Hackett Group, 2013 What began as a family-run sugar farm in Cuba in 500,000 tons per year the 1850s is now the largest sugar refiner in the could quickly deliver the real-time capabilities Annual Revenues: $5.5 billion (Fanjul Corp.) world. The company is involved in planting, necessary for increased flexibility and better Employees: 2,000 harvesting, processing, packaging, marketing and decision-making. www.floridacrystals.com distribution—all within its own farms and facilities. Source: Florida Crystals Recently, Florida Crystals expanded operations to A Way to Stay Ahead Europe and Latin America via acquisitions and Florida Crystals likes to be out front. More than a investment. It is also looking at opportunities in decade ago, it was the first to mill certified organic China and Africa. sugar to be sold in the United States. That has been true of its approach to using information To improve agility in its competitive cost- technology, too. The company has always situated

1. International Sugar and conscious industry, as well as to better manage itself on the leading edge of IT, from being among Sweetener Report. “Raw sugar its broad range of business activities and the first to roll out an ERP system in 1995 to its futures fall below 15 cents/lb for the first time since June 2010.” aggressive growth strategy, Florida Crystals’ early adoption of cloud computing when it rid itself Informa, Jan. 23, 2014. http://goo.gl/OFuOKu. leaders looked to information technology that of its datacenter in 2010.

Florida Crystals’ But in 2013, Florida Crystals was like many companies FIGURE 2 Better Decisions Through Data Challenges struggling with the limitations of its existing applications Business drivers for investment in analytics } Standardize and simplify and analytics environment: It was running numerous include the following. (% responding) infrastructure and batch processes that delayed data delivery and analysis, Improving the quality of decision-making applications. and it had devised workarounds that further complicated 59% } Create agile processes already complex business processes. Also like others, to facilitate growth Making quicker decisions through acquisition. Florida Crystals lacked real-time visibility into key 53% } Deliver data and enable business information. A study of global companies by analysis in real time. The Hackett Group found that fewer than half have Improving planning and forecasting 47% real-time data about customers, suppliers or financial Florida Crystals’ performance (see Figure 1, “Real Time vs. Reality”). Developing new products, services and Solutions revenue streams Furthermore, more than 40 percent of IT managers 47% } Deploy an in-memory suite of enterprise responding to a recent IDC survey say that users Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX applications to speed could not perform predictive analysis or work with information delivery to real-time data because they have to wait between from re-imagining business processes to take advantage of business users. two and 10 days for their data to be processed.2 it, noted business transformation expert Behnam Tabrizi, } Move mission-critical consulting professor at Stanford University’s department of business applications to a managed cloud service Managers at Florida Crystals used to spend much of management science and engineering. with flexible capacity their time waiting for reports to run or devising their to support geographic own spreadsheet analyses. But Vice President and No More Waiting for Data expansion. CIO Don Whittington chose to move the company’s It took two months for Florida Crystals to migrate its full } Use real-time analytics most important business systems to an in-memory stack of ERP software, from finance and materials to enable employees to platform that could deliver data to employees much management to sales and distribution, to a managed make better operational 3 decisions. more quickly than overnight batch processing. He cloud service. Then the company switched over to the also chose to run those systems using a managed in-memory platform during a long weekend, keeping cloud service with the capacity to process petabytes terabytes of data in sync with no disruption to the of data. The flexible capacity would accommodate thousands of employees who use that data in three the company’s expansion plans more cost effectively. countries on a 24-hour basis. The migration began late But Florida Crystals’ goal goes beyond merely on a Friday and, Whittington said, “it was literally producing reports faster. Ultimately, business leaders business as usual on Monday.” plan to transform the way the company operates. Only it was more than business as usual. Key With in-memory computing in the cloud, business performance processes such as accounts payable users can process reports on-demand, on their own analysis doubled in speed. Cost center reports that used and in real time—without waiting for IT to provide to take minutes took milliseconds. Programs that ran for data. Employees can now devote their days to the hours finished in minutes. End-of-month workloads that analysis and business decision-making that will took three days to complete now run in about four hours. underpin the company’s growth and differentiation In fact, response times for most business transactions 2. Olofson, Carl W. and Henry D. strategy. As batch jobs are replaced by real-time Morris. “Blending Transactions and have improved by anywhere from 50 percent to 500 Analytics in a Single In-Memory processing, the company can get information from its percent, according to Whittington. “We don’t just make Platform: Key to the Real-Time Enterprise.” IDC, February 2013. supply chain faster than ever before, and thus get work faster,” he said. “Work goes away.” http://goo.gl/e0vZP3. answers to customers more quickly. 3. SAP Technology. “Florida Crystals Speeds Up Supply Chain When employees do not have to download data to a With Enterprise HANA Cloud.” Eliminating batch processes is only the beginning, spreadsheet, format it, upload it again and then YouTube, July 8, 2013. http://goo.gl/1KVrEy. however. The greatest benefits from real-time data come reconcile it with the four or five other reports their

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Benefits of Real- colleagues have created, they have more time to FIGURE 3 Real-Time Businesses Time Business at study the data and use it to make decisions. Outperform Peers Florida Crystals Research by Erik Brynjolfsson, director of the MIT Compared to other companies in their industries, firms using advanced analytics are: } Self-service: Business Center for Digital Business, and his colleagues, found users can process that among companies using data in their decision- 200% as likely to be in the top quartile of financial reports on demand, making, productivity increases by 5 percent to 6 performance without going through IT. 4 } percent. Meanwhile, a survey of IT decision-makers Faster analysis: Most 300% as likely to execute decisions as intended business transactions by publisher IDG Enterprise found that companies are are 50 percent to 500 investing in analytics primarily to improve the quality 500% as likely to make decisions much faster than percent quicker, so and speed of their decision-making (see Figure 2, market peers business users have “Better Decisions Through Data”). more time to study Source: Bain and Company Insights, 2013. http://goo.gl/lY1IvA their data. } Improved flexibility A Real-Time Transformation There have been benefits for IT, as well, in line with Florida to support growth: Now, Florida Crystals’ executives are on the way to Crystals’ cost-cutting goals. Total cost of ownership for A scalable cloud transforming the company into a real-time computing and storage is down by an average of 30 infrastructure business. “Your day is no longer spent running percent. And by embracing the cloud, Whittington’s helps mergers and transactions,” Whittington explained. “It’s [spent technology group does not need dedicated resources or acquisitions go more taking] the results of those transactions and [using] expertise to manage the in-memory platform. That will smoothly and reduces IT costs. that to a business advantage.” “empower IT groups to become that part of the business } New value creation: and not worry about the technology,” he said. “We leave With access to real-time When it comes to implementing new technologies, that to our partners.” data, Florida Crystals such as in-memory and cloud computing, said can focus more on Narendra Mulani, senior managing director at And the more flexible systems will make the technology value-added work, Accenture Analytics, “the greatest value will come transitions for future corporate acquisitions and mergers such as improving inventory management; from the ability to enable decision-making at the go more smoothly. In the past, some of the acquired developing new right time.” companies did not come with their technology formulations, products applications, requiring Florida Crystals to get off the or packaging; or It is not so much that such new business systems are seller’s technology in as little as 90 days. “A flexible planning expansion into simpler, said Robert Parker, vice president of research analytic model and underpinning technology can new markets. at IDC Manufacturing Insights; it is that they are more actually be hugely beneficial by allowing the company to flexible. The same core technology can be used to evaluate performance even if systems aren’t fully meet varied and changing business demands, converted yet,” IDC Manufacturing Insights’ Parker said. without having to reprogram them constantly. “These approaches create the necessary resiliency that That should give Florida Crystals a leg up in its industry. allows a company to be more responsive to market A 2013 global survey of 400 large companies by Bain & conditions,” he said. They enable more data-driven Company found that firms using advanced analytics are decision-making at the strategic level and greater more likely to be in the top quartile of financial agility when responding to events in real time. performance within their industries, to execute decisions as intended, to use data very frequently when making 4. Brynjolfsson, Erik, Lorin Hitt and In a highly competitive and constrained market, decisions and to make decisions faster than market Kim Heekyun. “How Does Data- Driven Decisionmaking Affect Firm having flexible technology means Florida Crystals peers5 (see Figure 3, “Real-Time Businesses Outperform Performance?” 2012. http://goo.gl/Wkav0C and http:// can focus on value-added activities that will Peers”). Said Whittington, “It’s a game changer.” • goo.gl/9VUMG6. distinguish it from the rest, whether it is better Stephanie Overby is a freelance business and 5. Pearson, Travis and Rasmus management of inventory; new formulations, Wegener. “Big Data: The technology writer based in Boston. Organizational Challenge.” Bain and products or packaging; or planned expansion into Company Insights, Sept. 11, 2013. http://goo.gl/lY1IvA. emerging markets. This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Globus at a glance Globus is the premier department store subsidiary of The Federation of Migros Cooperative, the largest retailer and employer in Switzerland. The cooperative has more than 40 business enterprises that are divided into five strategic business units: Cooperative Retailing, Commerce, Industry & Wholesale, Financial Services and Travel

Industry: retail

2012 revenue: $849 million

Headquarters: The Spreitenbach, near Zürich, Switzerland

Products: high-end fashion, beauty, living and gourmet products

Employees: more than 3,000 www.globus.ch/en/index.html

Source: Globus

Page 48 RESEARCH UPDATE | Real-Time Business Case Studies

Globus Is Always in Style with In-Memory Technology Real-time insights enable Switzerland’s high-fashion department store to keep up with fast-changing consumer preferences and optimize its supply chain. BY JOE MULLICH he Federation of Migros Cooperative, the largest retailer and employer in Globus at a Glance T Switzerland, has long been noted for its supply chain excellence and creative } Industry: Retail thinking about logistics. Last year, for example, the company was honored with a } Company description: prestigious Swiss Logistics Award for determining the best transport routes for its Globus is the premier department store subsidiary of The Federation goods. It is not surprising that a company recognized FIGURE 1 of Migros Cooperative, Ready for Mobile Offerings the largest retailer and as the most innovative retailer in Switzerland, Respondents were asked which of the following employer in Switzerland. according to a survey by the University of St. they would use if offered on their mobile device The cooperative has Gallen, would also be a leader in finding better while in-store. more than 40 business enterprises that are ways to leverage data to improve its supply chain Gather loyalty points or savings as you shop, divided into five strategic and merchandizing techniques.2 based on current promotions business units: Coopera- 88% tive Retailing, Commerce, Industry & Wholesale, A prime example is Globus, Migros Cooperative’s Real-time promotions 88% Financial Services and premier department store subsidiary. Founded in Travel 1907, Globus operates 14 big retail stores that sell Order out-of-stock items in-store for home } Annual Revenue: 2012 clothing, cosmetics, jewelry, household supplies and delivery or put items on a wish list gross revenue was $849 82% million using current other upscale goods. The retailer’s need for continual 1 exchange rates supply chain improvement is driven by higher Shopping list, item locator and store navigator 79% } Headquarters: The production costs, a weak Euro and rising raw material Spreitenbach, near prices. Consumers, meanwhile, are more demanding Scan products as you put them in your Zürich, Switzerland and harder to please, which requires Globus to stay shopping cart } Products: High-end 77% on top of retail and fashion trends that can change as fashion, beauty, living and gourmet products fast as someone posts a new YouTube video. Base: 6,000 consumers from eight countries Source: “Accenture Seamless Retail Study,” April 2013 } Employees: More than 3,000 Like many worldwide retailers, Globus operates http://www.globus.ch/en/ in a competitive landscape, in which consumer sales performance data to identify merchandise index.html expectations are increasingly influenced by mobile that is selling slowly. Such goods are a costly

Source: Globus device capabilities such as comparing competitive issue for many retailers. Studies have shown the offerings and sharing their opinions on social media cost of owning excess inventory is approximately (see Figure 1, “Ready for Mobile Offerings”). 2.5 percent per month, or 30 percent per year, according to consulting firm The Retail Management Insights at the Shelf Level Advisors. Costs include interest, insurance, buying With this relentless pressure, Globus depends on expenses, receiving department expenses, property 1. Globus Web site. http://tinyurl.com/myfoz4k timely, accurate reporting and analysis of sales taxes, markdowns and shrinkage, the firm says.3 2. Migros Web site. http://tinyurl.com/m3f5o7s performance data to identify consumer trends so Another issue is customer satisfaction—if store shelves 3. Carter, Linda. “What Is Stock Turn Rate, And Just Why Is It So Important?” The it can change its pricing and marketing strategies contain too many items that customers do not want, Retail Management Advisors. http://tinyurl.com/mg5urrh as the market changes. In particular, Globus uses that leaves less room for the goods they do want.

Globus needed to find a way to track slow-sellers FIGURE 2 Speeding Insights with so it could then discount or promote those products In-Memory Computing to clear precious shelf space for hot products. Globus has seen quantifiable speed benefits from its Enabling this type of analysis was no easy matter, in-memory computing system. considering the large data volumes involved. Globus Before After

offers roughly 800,000 products through a network Data 550 gigabytes 60 gigabytes Globus’ Data of some 3,500 suppliers. With increasing sales warehouse (due to data Analytics volumes and transaction data, the amount of time it capacity compression) Challenges took Globus to generate reports used for inventory Time to 22 minutes } The 17 seconds Speed reporting. decisions was spiraling out of control.4 retailer wanted to generate a (reports did (including all generate reports more report on not include all product lines) quickly so it could Analytics at the Speed of Business slow-selling product lines) stay on top of rapidly items changing fashion trends. To get quicker insights, Globus implemented } Minimize excess inventory: in-memory database technology. By leveraging Time to 7 minutes 60 seconds Globus wanted to in-memory computing, Globus can now instantly generate a reduce the amount of sales report explore and analyze data from numerous sources. slow-selling inventory to make more shelf space Operational data is captured in-memory as Source: IBM for hot items. transactions occur, permitting analytics to be } Improve efficiency: The performed against rich data sources and providing retailer wanted to reduce the volume of data it the type of real-time insights that were simply not stored while increasing possible before. Sales promotion reports, which once took seven the insights and minutes to complete, can be produced in a mere information that data provided. This approach has provided numerous benefits. 60 seconds. “Previously, if we needed sales For example, Globus used to store 550 gigabytes figures of our 14 big stores, [we] had to generate Elements of Globus’ of data in its data warehouse. Thanks to its in- 14 separate reports and then combine the results Analytics Solution memory approach, that has decreased to only manually,” Weiss says. “Now we can simply } An in-memory create one report at the touch of a button.” He technology-based 60 gigabytes, one-ninth the original size. “This is analytics platform important, because the less memory we need, the adds, “we can now process more complex and that helps to quickly lower our licensing fees and maintenance costs,” comprehensive requests within seconds, providing generate insights, resulting in better says Alexander Weiss, team leader, processes and a range of new insights that were simply not demand planning, business warehouse at the company.5 available without in-memory processing.”9 merchandizing and marketing in real time. Queries and even forecasts that call up archive data The insights culled from this approach have are now faster than ever, dramatically improving the increased employee readiness to embrace speed of insights6 (see Figure 2, “Speeding Insights analytics. Before the use of in-memory technology, with In-Memory Computing”). “Previously, it took “acceptance of the company’s key performance 22 minutes to generate a basic slow-seller report indicator dashboard was low among managers,” that did not include the whole product line,” Weiss Weiss says. “With the new solution, the figures are says. “Access to these reports had to be planned referred to and used regularly, helping the company in advance because of the processing time.”7 But to manage its retail business more efficiently.”10 after implementing the in-memory database system, Globus can generate a slow-seller report for the entire All this leads to better inventory decisions, a product line in just 17 seconds. Employees obtain stronger bottom line—and happier customers. • 4., 5., 6. SAP Web site. http://tinyurl.com/lqvgz4k needed information 98 percent faster, eliminating the 7., 8., 9., 10. IBM. “Swiss Retailer Migros Joe Mullich is a freelance business and technology Spots Consumer Trends in Real Time with wait time by two hours per employee per day, so SAP HANA.” March 21, 2013. writer based in Sherman Oaks, Calif. 8 http://tinyurl.com/lfvruwk decisions can be made on up-to-the-minute data.

2 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Hewlett-Packard at a glance Provides hardware, software and services to consumers, small and midsize businesses, and large enterprises, including customers in the government, health and education industries Industry: high tech

Products: computer hardware and software; IT services; IT consulting

Founded: 1939

Headquartered: Palo Alto, Calif.

Sales: $112 billion

Employees: 331,800

www.hp.com

Source: Hewlett-Packard

Page 51 RESEARCH UPDATE | Real-Time Enterprise Case Studies

HP Paves the Future with Real-Time Insights For the high-tech giant, in-memory-enabled speed is a business game changer. BY JOE MULLICH or the past several years, the high-tech industry has faced the challenges of Frising customer demand, shortened product lifecycles and constant innovation. Now that technology has become a fundamental element of not just the workplace but also everyday consumer life, however, the pressure to execute quickly against these traditional challenges has never been greater.

Succeeding in this pressurized environment FIGURE 1 Big Returns From Big Data requires high-tech businesses to acquire real- High tech expects above-average Hewlett-Packard returns on big data investments. at a Glance time intelligence—about customer tastes and preferences, market conditions, supply and Utilities } Products: computer

demand—and to take immediate action. For all of 73.0% hardware and these reasons, the high-tech industry has become Energy & resources software; IT services; one of the leading users of big data and analytics, 60.6% IT consulting gleaning insights from the huge volumes of data High tech } Founded: 1939 produced internally, by the business, as well as 52.4% } Headquartered: externally, from its customers and partners. Palo Alto, Calif. Average 45.5% (Mean percentage } Sales: $112 billion of expected ROI Technology firms are seeing a payoff from their for 2012 big data } Employees: 331,800 Banking/financial services investments.) initiatives. In a study by Tata Consultancy Services, 43.7% www.hp.com high tech was among the top four industries in Source: Hewlett-Packard Insurance terms of spending on big data and in the top three 38.9% for expected return on investment (see Figure 1, Travel/hospitality/airlines “Big Returns From Big Data”). 37.9% Telecommunications Hewlett-Packard, the $112 billion technology giant, 37.9% is a leading proponent of the value of big data. “We Retail eat, sleep and drink very large databases and big 36.4% data as a present reality,” said Dave Carlisle, chief Life sciences architect and strategist for HP’s Enterprise Group.1 35.3% “We are under increasing pressure. Business Manufacturing expectations—and particularly user experience 28.9% expectations—are increasing significantly.” Source: Tata Consultancy Services. “The Emerging Big Returns From Big Data.” 2013 Consider, for instance, HP’s supply chain. Every 60 seconds, HP delivers 120 PCs, 100 printers, servers. Every one of those shipments is the 1,200 ink and toner cartridges, seven network result of countless actions and decisions made 1. Dave Carlisle spoke recently in a recorded Webcast. devices, seven terabytes of storage and five throughout the global supply chain. The question

FIGURE 2 In-Memory Initiatives HP is focused on three strategic initiatives, powered by in-memory computing. Strategic Initiative Business Innovation Data Consumption

Financial planning • Drive accelerated and sophisticated scenario 1–2 terabytes of data and 400 planning in real time business intelligence users • Engage users with instantaneous results when executing simulations and what-if scenarios

Quote-to-cash • Accelerate core ERP processes in key areas, 28 terabytes of data and 1,000 such as order management administration, business intelligence users month/quarter/fiscal year-end processing and live operational reporting, delivering modern “experience at scale” to users

Next-generation • Drive on-demand visibility and improved 7–10 terabytes of data; 1,500 business warehouse inventory leveling in the product supply chain business intelligence users; analytics • Enable near-real-time account visibility in the 3 predictive users global quote-to-cash process

Source: Hewlett-Packard

Hewlett-Packard’s the tech giant faces is, how to coordinate all those After running numerous tests, HP discovered query Business decisions to optimize pricing, reduce out-of-stocks speeds had improved from an average of two hours Challenges and shift manufacturing to align with customer using traditional databases to just 88.23 seconds using } Respond quickly to demand. “To succeed, organizations need to in-memory computing. As a result, HP employees will changing conditions via make fast, informed decisions by extracting large no longer experience radical swings in processing times fast analysis of real-time amounts of raw data from their systems to turn it that vary depending on where data is stored. With user- information driven, powerful analysis and simulation capabilities, HP } Enable employees to into information that is meaningful,” said Bill Veghte, analyze data quickly and chief operating officer at HP.2 workers will obtain immediate answers to virtually any intuitively business question—with minimal help from IT. } Optimize the use of data To succeed in this endeavor, HP is embracing volumes generated by in-memory computing, which enables real-time The increased speed will have a wide-ranging impact the supply chain decisions through instantaneous analysis of big on employee satisfaction and on meeting market demands data sets. In-memory computing will enable HP to and customers’ changing expectations. With built-in take immediate action based on up-to-the-second analytics, dashboards and reporting tools, data can be insights into its financial planning, ERP and inventory digested and presented intuitively, enabling employees to processes (see Figure 2, “In-Memory Initiatives”). quickly take corrective action and seize opportunities.

From Hours to Seconds “At the end of the day, speed is a business enabler,” The ability to analyze massive amounts of fast- Carlisle said. “It’s cool from a technical standpoint changing data in real time provides numerous benefits, that a report that used to run in three hours can now according to Carlisle. What HP employees will really be done in 30 seconds. But the key is using speed to notice is the unprecedented speed of accessing rethink business experiences and processes in ways 2. Hewlett-Packard. “HP Opens information to make decisions, he said, particularly not possible before.” For instance, employees will gain New Center of Excellence for In-Memory Computing.” HP press given the vast size of HP’s underlying databases and self-service access to financial information, enabling release, Jan. 10, 2013. http://goo.gl/P80n5J. the global scale of its multifaceted business. more rapid analysis and reporting capabilities and

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

significantly reducing the time needed for closing FIGURE 3 Order Management processes at the end of the month, quarter and Complexity fiscal year. For each process, respondents were asked how many IT systems are required to support it. One System Two Systems With better—and faster—visibility into financial, Three or More Systems inventory and supply-chain information, HP will Order management Hewlett-Packard’s be able to detect and respond to issues and 55% 21% 21% IT Solutions exceptions more quickly, in addition to proactively Demand management } Use in-memory managing risk and forecasting demand. HP could 53% 19% 15% computing to analyze potentially accelerate inventory turns and ensure vast amounts of data ERP profitable service delivery by instantly evaluating quickly 56% 9% 21% and making granular changes in supply and } Begin with a critical Warehouse management system involving small demand, reducing out-of-stocks and lost sales. 34% 21% 28% data sets } Develop next-generation Transforming processes and enabling new Price management business warehouse capabilities are at the core of what manufacturers 32% 15% 23% analytics are looking to achieve with analytics. “In our Production planning research, only 10 percent of manufacturing 32% 11% 30% companies are satisfied with their ‘what-if’ Tactical supply planning capabilities, and only 24 percent of companies can 36% 17% 15% easily determine the profitability of decisions,” said Transportation planning Lora Cecere, founder of Supply Chain Insights. 36% 9% 15% Manufacturing execution systems Tackling Order Management 23% 15% 23% HP’s in-memory computing efforts began with Product lifecycle management a focus on its largest single ERP environment, 34% 9% 11% which includes order management administration, month/quarter/fiscal year-end processing and live Source: Supply Chain Insights. “Go Big or Go Home?” July 2012. http://goo.gl/p3x8B operational reporting, accounting for over $41 billion in revenues. The HP Account Management Portal Throughout 2014, HP will use in-memory computing is used by thousands of HP order administrators to provide better data visualization, and it will begin worldwide. In-memory computing slashed the to incorporate mobile capabilities. Taking a staged time required to run a report from upwards of approach to in-memory computing enabled HP to seven minutes to 40 seconds. “Users can get their scale out its knowledge from both a technical and a display in real time while they talk with customers “comfort level” standpoint. “We are starting to deliver on the phone,” Carlisle said. What made this so disruptive capabilities now,” Carlisle said, “while compelling, he added, is that there was no need to understanding and planning for even larger scale retrain users, which increased the time to value. “It is changes over the coming years.” a good first step that brings material change without disruption.” With the combination of its real-time big data anlytics foundation and strong 2013 performance,3 HP seems Streamlining order management is a challenge for poised to do just that.”• many companies. According to a study by Supply Joe Mullich is a freelance business and technology Chain Insights, order management often requires 3. Meola, Andrew. “Why Hewlett- writer based in Sherman Oaks, Calif. Packard is Up Today.” The Street, more than one system (see Figure 3, “Order Jan. 13, 2014. http://goo.gl/v8OtZs. Management Complexity”). This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

HSE24 Germany, Austria, Switzerland at a glance A leading home-shopping network based in Germany

Industry: retail

Founded: 1995 2012 sales: €515 million

Employees: 725 at HSE24; approximately 2,100 external (call center, logistics) Customers: 8 million people have shopped at HSE24, equivalent to more than 10 percent of Germans over the age of 18

Products: 20,000 different products are sold each year Product segments: fashion (35 percent), beauty/wellness (23 percent), jewelry (21 percent) and home living/ household and DIY (21 percent) www.hse24.de

Source: HSE24

Page 55 RESEARCH UPDATE | Real-Time Big Data Analytics Case Studies

Using Real-Time Insights, HSE24 Gets Closer to Customers With in-memory computing and predictive analytics tools, Europe’s home-shopping network can offer a personalized experience—and solve costly issues—in real time. BY JOE MULLICH HSE24, a leading home-shopping network based in Germany, measures its business HSE24 Germany, in milliseconds. The 18-year-old company reaches 41 million households in Germa- Austria, ny, Austria and Switzerland, and it has now expanded into Italy and Russia. All of its Switzerland at a Glance home-shopping channnels must stay aware of customer buying patterns and behav- } Founded: 1995 iors and respond to them in real time. For example, if it appears that a sales spike is } Sales: €515 million depleting inventory after one of the channels (2012) FIGURE 1 Sharpening Insights in a displays a red shirt, it must immediately switch to } Employees: 725 at Digital Shopping World promoting a blue shirt to encourage purchases of Retailers understand the value of data but struggle to HSE24; approximately stock on-hand. turn it into profitable insights. (percent of respondents) 2,100 external (call Proprietary data can help differentiate the center, logistics) Considering that a new customer call comes retailer’s products and services } Customers: 8 million 67% through every two seconds, it is easy to people have shopped understand the company’s need for speed. at HSE24, equivalent Harvesting customer insight from data is a Retailers generally are moving to real-time key challenge to more than 10 strategies. But because home-shopping networks 58% percent of Germans are based on “selling live,” the growing trends over the age of 18 Data collection is a key challenge of big data analytics and real-time retailing are } Products: 20,000 52% particularly crucial for this retail niche. different products are Active users of analytics that extensively use their sold each year data to generate new ideas and opportunities “The amount of data available to retailers is } Product segments: 25% enormous,” said Don Peppers, an author and fashion (35 percent), Source: Accenture. “Seamless Analytics: Three Imperatives for the Retail founding partner of Peppers & Rogers Group, beauty/wellness (23 Digital Marketplace.” 2013. http://goo.gl/aFFy85. a marketing consultancy that specializes in percent), jewelry (21 customer-centric practices. Retailers are working immediate action on the insights gleaned. The percent) and home to find ways of using their data to improve approach enables the company to engage more living/household and operations and gain insights into what customers personally with its customers and create highly DIY (21 percent) want at the exact moment of need (see Figure 1, targeted marketing campaigns. Additionally, it “Sharpening Insights in a Digital Shopping World”). will help reduce product returns significantly— www.hse24.de Source: HSE24 an historically costly issue, particularly for As HSE24 continues its rapid expansion across omnichannel retailers. Europe—and fortifies its increasingly omnichannel strategy, including mobile, social networks and A Complex and Growing Business the Web—the company is turning to in-memory In the German-speaking market, HSE24 operates 18 computing as a foundation for success. This call centers and four logistics centers. It processes technology is enabling HSE24 to rapidly analyze orders for more than 1.5 million customers each year huge stores of customer information and take with strong partners such as DHL. For each hour of

live broadcasting, the retailer offers two to three new FIGURE 2 A Day in the Life products that range in price from €20 to €150, on On any given day in 2013, HSE24 serviced thousands of customers. average, and up to €35,000. In 2013, the company fielded 14.6 million calls and shipped 11.2 million Calls Received Parcels Shipped packages (see Figure 2, “A Day in the Life”). (Daily) (Every Weekday)

“We are running seven days [a week], 365 days a year,” Average 40,000 37,000 said Christian Schnetzer, CRM team lead at HSE24.1 HSE24’s Peak 100,000 74,000 Challenges “We need our systems every minute, every day.” } Capture, analyze and Source: HSE24 respond to customer The challenge for HSE24 is developing deep data in real time. relationships with its customers to both attend to In-memory computing enables companies to } Build highly targeted and quickly analyze very large volumes of up-to-the- personalized marketing and even anticipate their needs, despite no face-to- campaigns. face interaction. Doing so is becoming even more second and fast-changing data from diverse } Provide marketers with complex as the company expands (see Figure 3, sources, including sales and customer data from do-it-yourself tools to “Geographic and Channel Expansion”). internal systems and social media sentiment from reduce IT overhead. the Web. As such, in-memory computing seemed } Reduce costly product HSE24 increasingly interacts with customers on like an ideal way for HSE24 to keep pace with returns. the increased complexity of doing business by } Maintain a consistent digital channels, including online, mobile apps customer experience and social media. In fact, its electronic and mobile gleaning insights from its growing data volumes in across channels. commerce segments have grown at nearly twice the real time. rate of the company’s business as a whole. Because nontechnical users can ask complex These developments have caused an exponential questions of current data and receive answers in increase in customer data, as well as the need real time, they can take meaningful action during to keep up with dramatic shifts in taste from one small windows of opportunity, when the chance of country to the next. Michael Kuenzl, senior vice success is greatest. Additionally, these users can president of IT systems at HSE24, described obtain a complete picture of customer desires, the channel expansion as moving from “simple” enabling them to provide a personalized experience multichannel to “complex” multichannel. “They are and avoid costly errors in how they sell or market. a community buying with us and open to sharing “More and more, the value of big data is finding an their experience with the community,” whether on immediate reaction,” Peppers said. “Retailers want Facebook, Twitter or YouTube, he said.2 to be living in the ‘now’ moment.”

The “Now” Moment HSE24 applied in-memory computing to some of its It is now critical for HSE24 to monitor customers’ most pressing challenges in campaign optimization opinions, likes and dislikes, because the “digital and product returns. Before, when the marketing footprints” that Web shoppers leave behind can department wanted to target the customer profoundly influence marketing, inventory and segments that would be most likely to respond to customer service. Not to mention customers specific campaigns, it would need the business themselves, who express comments on social media intelligence group to analyze this information. That 1. Christian Schnetzer spoke at and expect service representatives to listen and process could take a week or more. a July 23, 2013, SAP Webinar titled “HSE24 Runs SAP HANA.” respond—quickly. It is fast becoming a necessity to http://goo.gl/jTgTNq Now, using customer engagement intelligence 2. Michael Kuenzl spoke at a develop an intimate understanding of customers’ July 23, 2013, SAP Webinar diverse and changing tastes, as well as to act applications running on in-memory computing, titled “HSE24 Runs SAP HANA.” http://goo.gl/jTgTNq instantly on these insights. marketers can perform the analysis themselves.

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Through an intuitive user interface, they can query FIGURE 3 Geographic and Channel the data in real time to detect customer buying Expansion patterns based on granular demographics, such HSE24 has grown by establishing businesses in new countries, new TV channels and new digital platforms. as age or where they live, Kuenzl said. New Business Launch The tools represent the analysis in graphics and Germany October 1995 charts, making it easy to see data patterns. Austria October 1996 HSE24’s Solutions Marketing staff can now drill down into millions } Implement a new of customer records in seconds, using recent Switzerland December 1997 in-memory customer data from a wide variety of sources. “We can do 16 hours of live May 1998 engagement intelligence the selection in seconds, and we don’t need the shopping each day solution. business intelligence department,” Kuenzl said. } Provide marketers with Online store December 1998 “We can play with the data.” intuitive and user-friendly customer segmentation HSE24 Digital (renamed September 2005 tools. Reducing Costly Returns HSE24 EXTRA) as a } Implement predictive HSE24 is also looking to in-memory computing second TV station analytics to uncover to reduce return rates and improve margins. HSE24 TREND “lifestyle” September 2010 hidden customer trends. According to global management consultancy TV channel Kurt Salmon, the cost of processing a return Shopping app for November 2010 can be two or three times that of an outbound smartphones shipment of the same item.3 The National Retail Italy June 2011 Federation, meanwhile, found that 40 percent of buyers intend to return a product when they order Russia April 2012 it, 40 percent order more variations of a product Shopping app for iPad November 2011 and 40 percent of all product returns are due to Shopping app on June 2013 poor product information.4 Google TV

Source: HSE24 For Web and phone- or mail-order companies, the freight and handling costs of returned goods are Once it detected this pattern, HSE24 could take particularly high. For HSE24, reducing returns by appropriate action. For example, it could improve one percent can lead to a seven-digit improvement its product descriptions to help customers more in the bottom line. Given these high stakes, the accurately select the right products on the very first company demonstrated how it could study clusters order. It also could develop a promotional campaign of customers who exhibit similar returns behavior to reward high-return customers if they lower their with the goal of reducing return rates. return rates over a specific period of time.

In the demo, HSE24 identified different groups of Best of all, Kuenzl said, marketers can e-mail customers who frequently return products. Using information to campaign managers so insights can be predictive analytics and in-memory computing, translated into action immediately. This is something it analyzed a particular group based on two HSE24 expects to do frequently in the future, as

3. Morrell, Liz. “The Value of elements: how many products they order and it explores more real-time uses for in-memory Returns: What Does it Mean for how quickly they pay for those products. It soon computing and its ever-growing mounds of data. • your Business?” RetailWeek, Feb. 19, 2013. became clear that these customers tend to http://goo.gl/4SPGz1 Joe Mullich is a freelance business and technology 4. Informatica. “How Rich Data order products in a variety of sizes or colors to Sells in Multichannel Commerce.” writer based in Sherman Oaks, Calif. On-demand Webinar. determine which they want to keep and which http://goo.gl/6jZkog they would return. This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Johnsonville Sausage at a glance Family owned company produces fresh, precooked and smoked sausage

Industry: consumer products

Headquarters: Sheboygan Falls, Wis.

Founded: 1945, family owned, privately held

Employees: 1,400

Presence: 6 manufacturing facilities in the United States; products available across the U.S. and in 30 other countries; sales offices in Canada, China, Japan and Mexico

www.johnsonville.com

Source: Johnsonville Sausage

Page 59 RESEARCH UPDATE | Real-Time Business Case Studies

Johnsonville Cooks Up a Real-Time Recipe Real-time business enables the nation’s leading sausage brand to continuously innovate, optimize performance and plan for the future.

BY MICHAEL S. GOLDBERG nnovation and consumer products go hand-in-hand. In today’s global Imarketplace, new offers, products and services need to be well-targeted, and they need to be implemented quickly. That is the mindset that drives Johnsonville Sausage at a Johnsonville Sausage, a family-owned enterprise promotions, according to a recent study by AMG Glance that has built a leading brand of sausage products in Strategic Advisors.1 Immediate insights from analytics Industry: Fresh, precooked the United States, to invest in real-time analytics. would enable Johnsonville to track special offers, such and smoked sausage as coupons. With real-time data, Johnsonville could Headquarters: Sheboygan To become more agile and ready to optimize correlate the offers with sales performance and adjust Falls, Wis. Founded: 1945, family- opportunities, the company is working to shift toward the programs based on those insights. owned, privately held an analytics-driven culture, says Ron Gilson, CIO and Employees: 1,400 vice president at Johnsonville. The goal, he says, is to The IT foundation for Johnsonville’s analytics Presence: 6 manufacturing embed real-time data-driven decision-making capabilities, which will be built over multiple years, is facilities in the United throughout the enterprise to anticipate future business based on real-time processing, enabled by in-memory States; products available needs rather than just respond to past events. computing. In-memory computing allows very large across the U.S. and in 30 other countries; sales volumes of data, from multiple sources, to be rapidly offices in Canada, China, One high-profile example is trade promotions. On and easily analyzed. The deployment also extends Japan and Mexico average, consumer products (CP) companies spend data visualization to mobile devices, to maximize www.johnsonville.com 13.7 percent of their gross sales on trade end-user interaction. But the strategy goes well Source: Johnsonville Sausage FIGURE 1 Most Consumer Products Companies in Early Analytics Stages (adoption rates of new technologies, %)

Completed Implementing Piloting or budgeting No plans 100 26% 4% 7% 4% 4% 11% 90 44% 33% 19% 45% 67% 80 55% 70 48% 60 44% 50 15% 41% 40 37% 30 20 22% 22% 22% 19% 1. AMG Strategic Advisors. 10 “The Trend Behind the Spend: 4% 7% A Study of Trade Promotion and 0 Merchandising Spending in the Software as a Digital media Mobile Social network Advanced Big data Consumer Packaged Goods (CPG) service (SaaS) platforms collaboration analytics Industry.” Acosta, 2012. Base: 27 leading CP companies http://goo.gl/Lnwhsg. Source: Grocery Manufacturers Association and The Boston Consulting Group, 2013. http://goo.gl/dFnylX

FIGURE 2 Analytics Vision A Quest for Growth Johnsonville Sausage expects to gain more visibility into its operations by collecting Johnsonville is pursuing its growth strategy and analyzing data in real time for all sources. Areas targeted for improvement include: from a position of market leadership in the Trade promotion management Track performance of retail promotions in real time, respond to United States. Three out of four bratwurst sold and optimization market trends, measure results and optimize outcomes. in the U.S. are Johnsonville. It’s also the No. 1 Integrated data repositories Provide sales, marketing, supply chain and other functions with a national brand of Italian sausage, smoked- consistent view of key performance indicators. cooked links and fresh breakfast sausage links. Mobile device dashboards Deliver and update results quickly to encourage users to explore Johnsonville’s reach expands across North data trends on devices such as tablets. America and to about 30 additional countries. Pricing Analyze retail point-of-sale activity in real time to optimize prices and maximize revenue based on demand fluctuations and supply. Even with its strong market posture, the Product launches Collect data on new product sales to analyze sales trends and anticipate future performance. company was spending 80 percent of its Supply chain management Gain real-time visibility into inventory, tracking movement from report-building efforts on preparing data for factory to store to consumer. analysis. Data was primarily historical in nature, Manufacturing operations Analyze shop floor data to predict when machines will need with limited forward-looking or predictive maintenance. capabilities. Shop floor data was collected but Source: Johnsonville Sausage not queried, analyzed or combined in an Benefits of beyond IT, as business leaders and managers efficient or standard manner. Promoting business In-Memory transition from a mindset of making gut-based intelligence reports using advanced data visualization Computing at decisions, based on decades of experience, to what was in its infancy. Viewed as a whole, the company’s Johnsonville Gilson calls “fact-based decisions.” analytics efforts were a collection of disconnected silos } Improved query that made it difficult to share information across performance time “We get new insights when we can put different } Positive executive business functions. response to tablet-based people with different perspectives in front of our visual data presentations data. That is what it’s all about: How can we An example is Johnsonville’s efforts to understand data } Ability to meet business generate new insights—because insights are the about market demand, Gilson says: “When we talk about objectives, such as more catalyst for innovation,” Gilson says. “That can be a demand signals, we have different languages. Marketing accurate sales forecasts new product innovation; it can be a new service. It speaks in marketing speak. Supply chain speaks in and marketing plans, could be a new process innovation within our optimized promotions, supply chain speak. Sales speaks in sales speak. We’ve visibility into competitive facilities or in our support groups.” got to get rid of that.” Unifying its approach to analytics posture, increased agility means creating an integrated and harmonized view of and responsiveness, and Throughout the CP industry, businesses are data, so everyone can examine the same data. increased profitability harnessing new IT capabilities like real-time big data analytics to pursue growth and innovation initiatives, The catalyst for investing in new IT capabilities was notes the Boston Consulting Group (BCG). In a 2013 Johnsonville’s need to optimize its trade promotions BCG study, 75 percent of CIOs at CP companies program. Timely tracking of its retail promotional activities said a top IT objective was supporting business is essential to managing its relationships with retailers, growth, compared with fewer than 20 percent in analyzing revenue and spending trends, and evaluating 2010. That said, the industry as a whole is still in the promotion effectiveness. This requires combining different early stages of incorporating analytics into data sets to derive insights about sales, correlated with operations, with most respondents piloting big data special offers like coupons and other retail promotions. and advanced analytics projects and few having Until recently, it has been an extremely manual and completed their implementations (see Figure 1, time-consuming effort to connect these dots in a way “Most Consumer Products Companies in Early that enables decision-makers to assess performance and Analytics Stages”). plan for the future.

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Johnsonville’s Gilson says the company chose systems that enabled FIGURE 3 Better Decisions Through Data Business it to modernize existing business intelligence and data Business drivers for investment in analytics Challenges warehouse systems with the in-memory platform and include the following. (% responding) } Optimize spending on relevant applications that collect and prepare data for Improving the quality of decision-making trade promotions by 59% analysis. Adding on to existing systems also enables measuring results and the IT staff to develop new skills in managing the Making quicker decisions making adjustments in 53% real time to maximize systems. Other business use cases add to the returns expected return on investment (see Figure 2, Improving planning and forecasting } Create an integrated “Analytics Vision”). 47% and harmonized view of information across Developing new products, services and Future Opportunities business functions revenue streams The first phase of the in-memory platform went live in 47% } Enable decision-makers May 2014, and Gilson says early indications are to assess performance Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX and plan for the future positive. He has seen performance improvements in query times, and the new tablet-based visual data processes to design, build and deliver products and Johnsonville’s presentations are popular among executives. services, Morris says. Solutions } Modernize existing Johnsonville plans to build on these early successes, business intelligence Picking Up Signals and data warehouse using a more centralized approach to analytics. Johnsonville exemplifies these trends by working to systems with real-time Establishing a harmonized data repository will enable speed the collection of market demand signals and analytics, enabled by an the company to track retail point-of-sale data, as develop truly real-time capabilities. It can analyze in-memory computing well as provide views into supply chain operations, product and promotion performance data by collecting platform such as shipments and inventory. Adding third-party market data from sales systems, retailers’ point-of-sale } Centralize data in a data sets, such as syndicated market data, will systems and syndicated research data. The analytics harmonized repository to track retail point-of- enable business analysts to gauge market trends. All platform lets Johnsonville test promotions, measure the sale data and provide of these activities are tied to critical business results and improve execution to maximize returns. views into supply chain objectives: improving sales forecasts and marketing “They can continually fine-tune things,” Morris says, operations plans; optimizing promotions to occur at the right “and make adjustments based on the outcome.” } Extend data visualization time and value; simplifying and integrating market to mobile devices analysis and Johnsonville’s competitive posture; With Johnsonville’s program underway, Gilson says the setting prices; increasing business agility and company believes in an “all-in” approach. “We’ve got to responsiveness; and increasing profitability. be able to be generating these insights daily, bring them through the innovation pipeline and execute on them. Efforts at Johnsonville to inject more and better data This is a critical point,” the CIO says. “It’s not about giving more quickly into business decision-making are an elite few the opportunity to do the analysis and do the emblematic of the real-time business and big data analytics and create insight. It’s about every member in analytics trend, says Henry Morris, senior vice Johnsonville on a daily basis, whether that is a small president of worldwide software and services at IDC insight driven through some predictive analytics for (see Figure 3, “Better Decisions Through Data”). maintenance, or some other transactional-oriented Companies across industries are pursuing strategies analytics, or it’s in sales and marketing going after the big to measure and improve the performance of wins. Really, it’s about orienting the entire organization’s business processes and to experiment with analytics on the future.” • opportunities to cut costs or drive revenue, using Michael S. Goldberg is a business and technology writer data to assess value and make changes in real time. and editor based in the Boston area. Data analytics is being targeted at financial processes, risk assessment, relationships and This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Kaeser Kompressoren SE at a glance One of the largest providers of compressed air systems and compressed air consulting services in the world

Industry: industrial machinery & components

Headquarters: Coburg, Bayern, Germany

Year founded: 1919

Number of employees: 4,800 worldwide

Countries of operation: 91

www.kaeser.com

Source: Kaeser Kompressoren SE

Page 63 RESEARCH UPDATE | Real-Time Business Case Studies

Kaeser Puts Customers First with Big Data and Real-Time Business The privately held German maker of air compressor systems and services taps predictive maintenance to drive ROI and a better experience for its customers.

BY LAUREN GIBBONS PAUL

Makers of heavy industrial goods aren’t exactly famous for their focus on the Kaeser customer experience. But a leading maker of compressed air systems is trying to Kompressoren SE at a Glance change that perception. Family-owned since 1919, Kaeser Kompressoren SE is Industry: Manufacturing and services solutions one of the largest providers of compressed air improve after-sales service. The company Headquarters: Coburg, systems and compressed air consulting services implemented a real-time business solution powered by Bayern, Germany an in-memory computing platform, which is able to Year founded: 1919 in the world. Its roughly 4,800 employees operate Number of employees: in nearly 100 countries across the globe. analyze vast amounts of granular, real-time data 4,800 worldwide produced by a machine-to-machine (M2M) interface. Countries of operation: 91 For a company that sells an industrial product in a This platform enables Kaeser to automatically monitor air www.kaeser.com business-to-business (B2B) environment, Kaeser compressors in use at customer sites. Kaeser is hardly Source: Kaeser Kompressoren SE is uncommonly focused on customers. The the only company using this technology: Companies are company’s Web site says its goal is to “provide forecast to step up their investments in M2M technology exceptional customer service coupled with over the medium term (see Figure 1, “Global M2M innovative products and progressive system Market Growth”). solutions.” Privately held Kaeser stands out in another way, as well, as the site makes clear: FIGURE 1 Global M2M Market Growth “You are doing business with a company with a Use of the technology that underlies predictive family tradition of producing quality equipment, maintenance, machine-to-machine communication, will expand along with maintenance analytics. not a company focused on meeting Wall Street (total annual revenues, M2M tech producers) estimates. Thomas Kaeser is proud to put his $21.52 billion name, his father’s name and his father’s father’s 2011 name on every product.”

$85.96 billion The Kaeser line of air compression systems 2017 includes machines many have never heard of:

rotary screw compressors, vacuum packages, Source: MarketsandMarkets.com. http://goo.gl/l1QTsO refrigerated and desiccant dryers and condensate management systems, as well as The Kaeser system is capable of running analytics portable compressors, filters and blowers. As against the vast amounts of M2M data generated from important as the quality of those products is the systems’ sensors and meters, including energy Kaeser’s commitment to after-sale service. consumption, operational status and compressed air quality. The objective: to predict which equipment will Predictive Maintenance need service, and when. Predictive analytics gives Powered by Big Data customers the luxury of planning downtime and In fact, Kaeser recently made a significant avoiding unexpected outages, which can cause investment to harness the power of big data to grievous injury to the bottom line.

Kaeser “Our customers simply can’t afford any FIGURE 2 Advanced Maintenance Analytics to Rise Kompressoren unplanned system downtime,” noted Kaeser ABI Research forecasts predictive and prescriptive Business Kompressoren AG CIO Falko Lameter in a maintenance analytics will dominate the analytics market within ﬁve years. (advanced maintenance analytics Challenges recent article.1 The company’s predictive percentage of total analytics market) } Make maintenance and maintenance ensures that its superior products other services offerings 23% stay that way even after years of use. “Our 2014 more cost-efficient and more valuable to products are built for a lifetime,” Lameter said. customers “But to get optimal longevity and performance, 60% } Streamline the supply they must be maintained properly.” Kaeser’s 2019 chain real-time predictive maintenance services help } Innovate through new customers achieve nearly 100 percent reliability. Source: ABI Research. http://goo.gl/9oJm67 technologies and business models Crucially, customers no longer have to monitor breakdown gets to occur, so over time both Kaeser their mission-critical air compressors around the parties can save money. Kompressoren Solutions clock. Kaeser handles that chore through its M2M } Implement a real-time interface, with resources on call to address issues “We’re always thinking of new ways to run things business solution swiftly. The system has direct, M2M connections better and innovate new technologies and business powered by an in- and real-time alerts, so potential failures can be models,” Lameter explained. Case in point: Kaeser’s memory computing detected before they occur, with corrective action transformation into a real-time business that puts the platform to enable undertaken before business is affected. customer at the center and focuses on delivering automatic monitoring solutions, not mere products. The services-based of customer site air compressors Lameter believes the addition of predictive analytics business model not only secures the company’s } Use predictive analytics for its equipment sets Kaeser apart from the market position and keeps it one step ahead of to help customers plan competition—a belief that is driving investment increasing customer demands, it also improves downtime and avoid across the business spectrum (see Figure 2, margins and revenues. In fact, total revenues from the unexpected outages “Advanced Maintenance Analytics to Rise”). sort of predictive analytics that Kaeser uses are } Use an M2M interface expected to grow sharply in coming years (see Figure to monitor customers’ mission-critical air ABI Research principal analyst Aapo Markkanen 3, “A Growing Market”). compressors around the says the cost savings associated with the ability clock, with resources on to predict equipment breakdowns before they occur Using real-time analytics, Kaeser can monitor call to address issues is potentially huge, both for Kaeser (or any other key parameters in real time, including power swiftly solutions manufacturer and service provider) consumption, operational availability and safety or } Establish a portal to and its customers. “From the manufacturer’s compressed air quality at each customer air station accelerate problem resolution and enable perspective, the key benefit is the ability to against minimum and maximum allowed values. customer service transform a product business into a product- Service engineers can analyze this real-time data from personnel to be more and-service business,” he says. a portal without having to visit the customer site. proactive and more Serving key data to the portal accelerates the customer-oriented “Many manufacturers have been trying to do this resolution of any problem, keeping customer for years, with their maintenance and other services operations reliable and efficient. With the offerings, but what makes the difference here is comprehensive view that the portal provides, that the combination of [M2M communication] and customer service personnel have become more analytics can make such services more cost- proactive and ever more customer-oriented. efficient from the vendor’s point of view, as well as more valuable from the customer’s point of view,” “What makes our predictive maintenance solution Markkanen says. Maintenance is conducted only cutting-edge is its computational complexity combined

1. http://goo.gl/VHzpWK when truly needed, he says, but still before a with the scale on which it operates,” wrote Lameter in

2 Bloomberg Businessweek ResearchResearch Services Services RESEARCH UPDATE | Real-Time Business Case Studies

2 Benefits of a recent blog post. “It collects thousands of FIGURE 3 A Growing Market In-Memory real-time data streams from customer air stations; ABI Research forecasts worldwide maintenance analytics revenues for the next ﬁve years will top Computing big data scales into the terabyte range. We $24 billion. (projected revenues from maintenance at Kaeser collect, analyze and act upon that huge amount of analytics, CAGR: 22%) Kompressoren data as part of our routine operations.” } Improved equipment $9.1 billion uptime 2014 } Decreased time to The predictive maintenance portal handles these problem resolution complexities while providing service engineers } Reduced operational with user-friendly, Web-based and mobile access. 2019 $24.7 billion risks As a result, the portal keeps Kaeser service } Accelerated innovation engineers fully informed of equipment status at Source: ABI Research. http://goo.gl/9oJm67 cycles customer sites so they can provide a } Better alignment between Kaeser’s consistently positive customer experience. and development of our next-generation products and services equipment,” Lameter wrote. and its customers’ The power and scale of the system provide a real needs advantage. “We had to harness the power of big This is an important benefit, Markkanen says. data to remain at the vanguard of our industry in “Manufacturers are able to study their products’ the 21st Century,” Lameter said recently. At the behavior in the field more closely than what has same time, the company simplified its software been possible before, so much of the analysis done landscape by migrating its CRM application to a for maintenance purposes can be used to steer the robust in-memory platform. The in-memory product development work, too,” he says. “If certain, platform helped Kaeser accelerate its core previously hidden circumstances prove detrimental business processes and achieve operational to the product time and again, it may be possible to excellence, he wrote. address them when designing the product.”

Running on the in-memory platform, database In this way, product development can become more response times are now five times faster than of an iterative process, with manufacturers before. “This has allowed us to bring the continuously learning about their products and entire lifecycle of the sales process under careful improving them accordingly. “All this can make a scrutiny—from lead management to requirements manufacturer like Kaeser more competitive in its analysis, solution planning and solution market,” Markkanen says. implementation. And with real-time information, we have streamlined our supply chain to deliver The big data and in-memory platform that form the on customers’ changing needs while generating basis of Kaeser’s predictive maintenance service have healthy margins,” Lameter wrote in the blog post. produced substantial benefits for the nearly 100-year- old company. According to Lameter, “We are seeing Aligning Product Development with improved uptime of equipment, decreased time to Customer Needs resolution, reduced operational risks and accelerated The predictive maintenance system has had a innovation cycles. Most importantly, we have been able significant, unforeseen side effect. Stored in to align our products and services more closely with Kaeser datacenters, the big data culled from our customers’ needs.”• monitoring machine operations forms the basis for Lauren Gibbons Paul has written extensively on sophisticated data analysis, including insights customer relationship management and customer relating to new product development. “This helps experience management for more than 15 years. us further optimize our service processes and This research project was funded by a grant from SAP. 2. http://goo.gl/iyjdoy furnishes highly valuable input for the research

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Maple Leaf Foods at a glance A leading CP company that specializes in meat and bakery products as well as agribusiness

Industry: consumer products

Headquarters: Toronto, Canada

Founded: 1991

2012 revenues: $4.9 billion (Canadian) Employees: 1,200

www.mapleleaf.ca

Source: Maple Leaf Foods

Page 67 RESEARCH UPDATE | Real-Time Business Case Studies

Maple Leaf Foods Turns Over a New

Leaf with Up-to-the-Minute Insights Real-time business enables Maple Leaf Foods to keep pace with fast-changing market conditions, increase profitability and hone a competitive edge.

BY JOE MULLICH

n a world that moves at Internet speed, consumer products (CP) companies Iface a particular challenge in responding to fast-changing tastes. And taste is literally the issue for Maple Leaf Foods, a leading CP company that specializes in Maple Leaf Foods meat and bakery products, headquartered in Toronto. Product pricing is also a at a Glance Industry: Bakery, meat, make-or-break issue, with ever-fluctuating FIGURE 1 Three Core Businesses agribusiness consumer demand and commodity prices. Maple Leaf Foods is organized into three Toronto, Headquarters: Because its primary products have extremely short major groups. Canada shelf lives (see Figure 1, “Three Core Businesses”), Meat Products Bakery Products Agribusiness Founded: 1991 Products Revenues (2012): $4.9 Maple Leaf needs to move swiftly to ensure its new billion (Canadian) product introductions, marketing and pricing Packaged Premium fresh, Hog production Employees: 1,200 strategies will translate into sales, profits and meats and frozen and meals specialty bakery www.mapleleaf.ca customer satisfaction. products Source: Maple Leaf Foods To meet these challenges, Maple Leaf needed a Fresh pork, Fresh pasta and Rendering and poultry sauces biodiesel real-time understanding of its complex business and turkey operations operations. Over the years, the company has products grown through dozens of acquisitions, leaving it Source: Maple Leaf Foods with a multitude of systems. That is why, about five years ago, Maple Leaf launched an initiative to Number One in Many Markets coordinate its diverse data volumes and is now Open a cupboard or refrigerator in the typical embracing in-memory computing—technology that Canadian home, and you are likely to find Maple Leaf, enables multiple sources of data to be rapidly and Schneiders, Dempster’s or one of the company’s other easily analyzed. brands. Maple Leaf Foods ranks number one in market share in many Canadian food categories, including The company’s efforts have resulted in integrated fresh bakery products, branded prepared meats, analytics across business functions and chilled ready-made meals, branded fresh poultry, and instantaneous insights to support real-time fresh pasta and sauce. Additionally, the company has decision-making. This is a critical issue for many operations in the United States, the United Kingdom, CP companies, which need to respond quickly to Asia and Mexico. To stay ahead of the competition, rapidly changing market conditions. “Ultimately, Maple Leaf introduces 100 or more new products the big piece is profitable growth so we can every year, such as its Maple Leaf Natural Selections, actually understand what’s happening in the a line of deli meats that contain no added marketplace quickly and secure more profitable preservatives, artificial ingredients, fillers or MSG. growth as a result,” said Michael Correa, vice 1. Canadian Financial Executives Research Foundation. “Real Time president, Information Solutions, at Maple Leaf Although an acquisition spree was key to Maple Leaf’s Business: Making The Most Of Big Data.” 2012. http://goo.gl/TujxFE. Foods.1 growth, it also left the company with 65 plants and 35

Benefits of Real- enterprise resource planning (ERP) systems. Because this meant consolidating its data into a single ERP Time Business at of the difficulty connecting information among these system, enabling integrated analytics across business Maple Leaf Foods isolated systems, Maple Leaf relied on weekly functions and developing the ability to analyze its data in } Standardized view summaries of business performance data, which real time. The first move in this direction began in 2008, across the enterprise: hampered its ability to respond to fluctuations in supply when Maple Leaf launched Project LeapFrog, in which it System consolidation and demand. In addition, business users were unable consolidated its many ERP systems and implemented and process standardization have led to easily drill down into reports to truly understand the common business processes across the organization. to improved operations supply and demand of individual products. through the use of “A big part of the business is protein—poultry, hog and in-memory analytics for These issues grew particularly acute between 2009 prepared meats—and each had their own way of doing management. and 2011, when the Canadian dollar appreciated work,” Hutchinson said. “And if you go over to the bakery } Faster analysis by 50 percent, resulting in Maple Leaf operating at side, like Dempster’s or frozen foods, each of them ran speeds: Business users 3 are experiencing speeds a cost gap of 15 percent to 25 percent versus its almost every process internally their own way.” that are 100 to 1,000 U.S. competitors. “If [Maple Leaf is] going to grow, times faster than with it can’t rely on the exchange of the U.S. dollar,” An important data source that Maple Leaf needed to traditional databases. said Dr. E. Jeffrey Hutchinson, global CIO of Maple integrate with its internal systems was the fluctuating } Self-service: Users can Leaf Foods. “If it’s going to compete against the market prices for “hogs and hams,” Correa said, which now run their own ad U.S. and European players, it needs the right can change throughout the day. Because the company hoc queries and reports. 2 trades such products on the mercantile exchange, he } Real-time pricing capabilities in house.” analysis: Maple Leaf said, “We’re looking at real-time, live information.” will be able to price Building the Analytics Foundation Maple Leaf needs to combine supply and demand products based on the As the competitive landscape toughened, the information from its transactional systems with data cost fluctuations of raw company realized it needed a better handle on its generated by the marketplace to enable real-time materials by combining cross-organizational performance. For Maple Leaf, pricing analysis. syndicated data with supply and demand information from FIGURE 2 transactional systems. Most CP Companies in Early Analytics Stages (adoption rates of new technologies, %) } Future use cases: The company is evaluating Completed Implementing Piloting or budgeting No plans the use of in-memory analytics for recipe 100 26% 4% 7% 4% 4% 11% development, demand 44% 33% 19% 90 45% planning, production 67% planning and trade 80 55% promotion to help it 70 48% further improve decisions 44% and profitability. 60 50 15% 41% 40 37% 30 22% 20 22% 22% 19%

2. Schick, Shane. “Maple Leaf 10 7% Foods’ ERP system project: A CIO 4% strategy.” IT World Canada, April 0 18, 2011. http://goo.gl/8daaTX. Software as a Digital media Mobile Social network Advanced Big data 3. Schick, Shane. “Maple Leaf Foods’ ERP system project: A CIO service (SaaS) platforms collaboration analytics strategy.” IT World Canada, April Base: 27 leading CP companies 18, 2011. http://goo.gl/8daaTX. Source: Grocery Manufacturers Association and The Boston Consulting Group, 2013. http://goo.gl/dFnylX

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Maple Leaf To enable real-time insights and provide employees FIGURE 3 Better Decisions Through Data Foods’ Business with a range of analytics tools, Maple Leaf Business drivers for investment in analytics Challenges embraced in-memory computing. Its first in-memory include the following. (% responding) } Real-time information project focused on customer data that contributes Improving the quality of decision-making and analytics were to profitability, such as profit and loss, price 59% required to optimize elasticity, and sales and marketing data, all of which product pricing and Making quicker decisions amounted to about one terabyte of data. Although enable other business 53% performance insights. Maple Leaf had used analytics for more than a } Data needed to be decade, in-memory computing enables the Improving planning and forecasting 47% consolidated from 35 company to ask entirely new questions; for example, ERP systems. “How should products be priced tomorrow in line Developing new products, services and } Processes needed to be with supply and demand?” revenue streams standardized across the 47% enterprise. } Business users needed The In-Memory Edge Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX the ability to drill down The ability to answer such questions through into reports. analytics puts Maple Leaf ahead of the pack. databases.5 For example, profit-and-loss analysis previously According to a late 2013 study by the Grocery required 15 to 18 minutes, but it now takes 15 to 18

Maple Leaf Foods’ 6 Solutions Manufacturers Association and The Boston seconds with in-memory computing, according to Correa. Consulting Group, only 23 percent of CP The in-memory computing system has also enabled a } Consolidated systems and standardized companies had implemented or completed an self-service data platform on which users can run ad hoc processes advanced analytics initiative (see Figure 2, “Most queries and reports (see Figure 3, “Better Decisions } In-memory technology CP Companies in Early Analytics Stages”). Through Data”). to provide instantaneous analysis of disparate data And according to Lora Cecere, head of research The use of in-memory computing has simplified Maple } Analysis and visualization firm Supply Chain Insights, only a small percentage Leaf’s business rules, processes and data, removing tools for hundreds of users of companies are satisfied with their “what-if” complexity and enabling collaboration. Business users capabilities and few can easily determine the can create more reports and queries, rather than profitability of decisions. “They need more real-time relying on the old system of canned reports. At the monitoring, dashboards, early-warning systems and same time, the company has achieved a standardized other imperatives that can only be accomplished view across the enterprise, enabling it to better through smarter use of analytics,” she said. manage its operations.

In Maple Leaf’s case, such insights could potentially As Maple Leaf begins using predictive analytics supported 4. Schick, Shane. “Maple Leaf guide the company on product introductions, as by in-memory computing, it plans to price products based Foods’ ERP system project: A CIO strategy.” IT World Canada, April well. “We’ll start to see what the reality is and be on the cost fluctuations of raw materials—such as wheat, 18, 2011. http://goo.gl/8daaTX. 5. Yuhanna, Noel, Holger Kisker, much faster at adjusting to market conditions as pork and chicken—and combine this information with Ph.D. and David Murphy. “Case they change,” Hutchinson said. “We’ll be able to syndicated data from sources like Nielsen to obtain Study: Maple Leaf Foods Relies On SAP HANA To Enable Faster make sure that our trade promotion monies, for real-time pricing information. The company is also Business Analytics.” Forrester Research, Oct. 28, 2013. example, are giving us what we want to get out of evaluating many new use cases, such as recipe http://goo.gl/QGBfXt. 6. Henschen, Doug. “Oracle them. Too often, unless you are very proactive there, development, demand planning, production planning and In-Memory Apps: Hunting Hana.” it just becomes a slush fund.”4 trade promotion to help it improve decisions and InformationWeek, April 11, 2013. http://goo.gl/nGvZns. profitability even further.7 • 7. Yuhanna, Noel, Holger Kisker, Ph.D. and David Murphy. “Case In early 2012, Maple Leaf extended its real-time Study: Maple Leaf Foods Relies Joe Mullich is a freelance business and technology writer analytics capabilities to 800 business users. In some On SAP HANA To Enable Faster based in Sherman Oakes, Calif. Business Analytics.” Forrester cases, business users experienced speeds that were Research, Oct. 28, 2013. http://goo.gl/QGBfXt. 100 to 1,000 times faster than with traditional This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Mercedes-AMG at a glance The performance unit of Mercedes-Benz, it tests the limits of driving dynamics and performance with hand-built engines

Industry: automotive

Parent company: Daimler AG

Headquarters: Affalterbach, Germany

Founded: 1967

Employees: approximately 900

www.mercedes-amg.com

Source: Mercedes-AMG

Page 71 C100 M94 Y0 K47 C85 M59 Y28 K28 C65 M29 Y79 K22 C33 M32 Y72 K3 C15 M72 Y100 K26 C48 M89 Y84 K49

CASE STUDY Talent & HR Management & Finance Technology Small/Midsize Marketing Business Operations Business & Sales Mercedes-AMG: A Showcase for Real-Time Business Decisions

C100 M94 Y0 K47 C77 M77 Y31 K16 C54 M84 Y20 K3 C85 M59 Y28 K28 C78The M52 Y28maker K6 C73 M19 Y40 K1 C65 M29ercedes-AMG Y79 K22 C61 M20 is Y78 the K3 performance Figure 1 Analytics O ers Relief From Operational Pressures of high- Munit of Mercedes-Benz, testing the Need to reduce the cost of manufacturing operations limits of driving dynamics and performance performance 70% with its hand-built engines. But the division automobiles is Need to improve delivery of new products to market building next- is also a testing ground for revving up 38% C41 M44 Y77 K15 C33 M32 Y72 K3 C15 M72 Y100 K26 C0 M60 Y100 K17 C28 M88 Y70 K19 C48 M89real-time Y84 K49 C61 business M66 Y61 K52 performance. Management of global supply chains generation 36% Management of demand volatility manufacturing As manufacturers integrate complex machinery 32% with networked sensors to create next-generation operations on Customer mandates for improved order fulfillment production processes, they are gathering more data a real-time 27% than ever before that could eliminate waste and Source: Aberdeen Group, “Big Data in Manufacturing Operations,” business improve efficiency, according to the U.S. National 2012. http://goo.gl/E43v5f Cutting costs is a top reason for seeking more 1 platform. Institute of Standards and Technology. In fact, insight from data. pressure to cut costs while bringing new products BY STEPHANIE to market is pushing manufacturers to make better OVERBY use of that data, the Aberdeen Group has found cars, such as its million-dollar S-class line and the (see Figure 1, “Analytics Offers Relief From new GLA compact SUVs, but also invested in more Operational Pressures”). efficient manufacturing processes to cut the cost of product delivery and improve vehicle quality. In the automotive industry, manufacturers must Daimler saw Mercedes-AMG as the perfect place to manage a greater number of both models and test out real-time systems that might benefit the customization options as they adjust to fragmented larger enterprise. consumer demand, while also managing shorter product life cycles, according to McKinsey and Employees working in the performance unit’s Company (see Figure 2, “More Product Options, engineering-focused manufacturing environment Shorter Product Cycles”). To make the best use of have yearned for real-time access to information for their data riches as they tackle these challenges, nearly 50 years, observes Dirk Zeller, Mercedes-AMG companies are implementing technology capable of head of IT Consulting. In addition, Mercedes-AMG analyzing large volumes of data from disparate operations are concentrated at the company’s 1. National Institute of Standards and Technology. “Real-Time sources in real time. Affalterbach, Germany, headquarters. “We have Data Analytics for Smart everything close to us, from design and development Manufacturing Systems Project.” http://goo.gl/KZ9nzW. Mercedes-AMG and its parent company, Daimler to marketing and sales, making us a good company AG, are among the businesses that are expanding to test out innovation, with regards to the products their product portfolios. Daimler is currently one of as well as in the IT area,” Zeller says. the largest producers of premium cars and the AUGUST 2014 biggest global manufacturer of commercial LESS WAITING, MORE WORKING © Copyright 2014. vehicles. To increase market share and profits, the In late 2012, Mercedes-AMG began to deploy an Forbes Insights. €118 billion company has not only introduced new in-memory platform across business functions to All rights reserved. FORBES INSIGHTS 1 FORBES INSIGHTS CASE STUDY

analyze large volumes of data in real time. The Figure 2 More Product Options, Shorter company started with back-office systems, such as Product Cycles accounting and finance, and then moved on to its Number of models Avreage product life cycle (months) core operations—the development and manufacture 131 122 111 110 of its one-of-a-kind vehicles. 106 100

Ultimately, The first goals were to “be more efficient by crunching numbers faster,” says Mercedes-AMG CIO Mercedes- 44 Reinhard Breyer. Company leaders saw an 32 30 AMG is 22 opportunity to free financial analysts from waiting 12 15 considering for reports and enable them to focus on strategic 2002 2011 2002 2011 2002 2011 simplification analysis and decision-making. BMW Audi Mercedes-Benz of the Source: McKinsey & Company, “Manufacturing the Future: The Next Era of Global Growth and Innovation,” 2012. http://goo.gl/EcQ6tq But a faster, more efficient back office was just a start. hundreds of A greater number of models and options and shorter systems it The key to becoming a real-time enterprise is not product life cycles challenge automakers. uses to delivering more data quickly, it’s more quickly delivering information that can improve the processes equipment, and efficient use of the facilities was a engineer, sell that lead to competitive advantage. For Mercedes- priority. The ability to correlate historical test data in and service AMG and Daimler, these are product development and real time with sensor data from the engines being cars to run on manufacturing processes. tested enables engineers to identify possible a single, problems more quickly. in-memory Real-time efforts “are useless if they aren’t aligned,” platform. says Behnam Tabrizi, consulting professor at “We are able to analyze test data from engines and Stanford University’s department of management test vehicles faster than before,” Zeller says. “And that science and engineering. “If you throw new leads to more insights faster, as we compare more technology at a bad process, you won’t resolve data and use complex analytics without losing time.” anything. It’s not just about real-time systems, it’s about real-time organizational transformation.” Now an engine test that shows unusual behavior can be stopped at any step of the test procedure; A NEW WAY TO TEST ENGINES engineers used to have to wait to analyze the majority Even today, every Mercedes-AMG engine is hand of data until after the hour-long test run has ended. built from start to finish by a master engine builder: That change, in turn, creates more engine-testing it’s a hallmark of the brand. But company leaders capacity each week. It also gives engineers more time recognized an opportunity to use big data analytics to focus on refining engine quality. to also improve the engine production process. When they went looking for the next high-value area A PLATFORM FOR REAL-TIME to transform the company using real-time TRANSFORMATION applications, they zeroed in on quality assurance What’s more, Mercedes-AMG now has a scalable for its engines. platform that it can apply to other areas of its business, and which Daimler can begin to apply Earlier this year, Mercedes-AMG piloted a real-time across the enterprise. Both Mercedes-AMG and quality assurance platform that harnesses predictive Daimler are expanding their vehicle lines, and thus analytics to optimize engine-testing processes when the number of projects they must manage. manufacturing its vehicles. The dynamometers the Mercedes-AMG has already introduced three new company uses to determine the torque or power compact models in pursuit of new market segments. characteristics of engines being tested are high-cost

2 FORBES INSIGHTS FORBES INSIGHTS CASE STUDY

The company is also enabling car buyers to FIGURE 3 Delivering Real-Time Supplier Data customize their vehicles more than ever before. Customer information The strategy seems to be working: in 2013 Mercedes 46% AMG sold more than 32,000 automobiles, its most Employee data successful year ever. 46% Product data “We are able to The project expansion and growth could strain 42% analyze test existing project management processes, so Supplier information Mercedes-AMG leaders have begun exploring how data from 23% engines and test the same data analysis and predictive functionality can be applied to engine testing for project Source: Accenture, “High Performance in IT: Defined By Digital,” 2013. vehicles http://goo.gl/7cc4KM management. The goal is to be able to visualize in dramatically Among top-performing companies, nearly half provide customer real time how the product creation and engineering and employee data to employees in the moment, but less than a quarter can do the same with supplier information. faster than process is progressing and to predict potential before.” bottlenecks, such as supplier delays, before they –Dirk Zeller, impact manufacturing ramp-up. who will be consuming the data and what they need to head of IT know. Business leaders and managers shouldn’t become Consulting, Both Mercedes-AMG and parent company Daimler overwhelmed by the amount of data available to them, Mercedes-AMG anticipate benefits beyond managing an individual Zeller advises. The concern is similar to that which project more efficiently. They want to take accompanied the introduction of email, Zeller recalls. advantage of real-time analytics to give employees “People felt disturbed by too many alerts, tending to who work across several projects, such as those with ignore them over time,” he says. “During our projects, a focus on braking systems, visibility across all we need to define if our target audience is prepared for projects, so they can see if a new part is in time, the solutions we’re offering.” quality and budget during the engineering process, if it is available for testing or inventories are low. Daimler’s IT group recently awarded Mercedes-AMG several innovation awards for the testing and Even among top-performing companies, access to implementation of its real-time systems. The parent such supplier intelligence in real time is relatively company will be using the performance unit’s rare. Less than one-quarter of leading enterprises experience and lessons learned as the basis for its can provide up-to-the-minute supplier data to their planned implementation of in-memory analytics. employees (see Figure 3, “Delivering Real-Time “They’re the trigger that started us on this path,” Zeller Supplier Data”). Mercedes-AMG, however, plans to says. “And we had a huge will to innovate.” • be the exception.

The goal is to have new business intelligence ABOUT FORBES INSIGHTS software running on the in-memory databases by Forbes Insights is the strategic research and thought leadership practice of Forbes Media, publisher of magazine and Forbes. the end of the year, and then to release new Forbes com, whose combined media properties reach nearly 50 million functionality every two months. “We are working business decision-makers worldwide on a monthly basis. very closely with our business partners to deliver Bruce Rogers exactly what is needed in an easy-to-use user CHIEF INSIGHTS OFFICER Brian McLeod experience,” Zeller says. DIRECTOR, NORTH AMERICA Writer: Stephanie Overby is a freelance business and technology One key to success when deploying real-time writer based in Boston. capabilities is gaining a deep understanding about

FORBES INSIGHTS 3 Real-Time Enterprises A Research Collection

National Football League at a Glance A professional U.S. football league that runs a

17-week regular season

Industry: sports & entertainment

Headquarters: New York, New York

Employees: 1,100

Founded: 1920

Number of Teams: 32

Viewership: 205 million people tuned in to NFL games in 2013

www.nfl.com

Source: National Football League

Page 75 RESEARCH UPDATE | Real-Time Business Case Studies

The NFL’s Advanced Analytics Score with Football Fans Looking to spur fan engagement, the National Football League is putting user-friendly data analytics tools in the hands of fantasy football fanatics.

BY STEPHANIE OVERBY rofessional football is America’s true national pastime. Viewership of NFL Pgames has grown steadily over the past decade. During the 2013 football season, NFL games attracted three times more eyeballs than other The National Football League primetime TV offerings, and 34 of the 35 most- FIGURE 1 NFL’s Gameday TV Viewership at a Glance watched TV shows in autumn 2013 were NFL games Gameday outdraws the leading primetime Industry: Sports and (see Figure 1, “NFL’s Gameday TV Viewership”). broadcast television shows. (average in millions of viewers) entertainment NFL broadcast Primetime broadcast NFL league office But if regular viewers of NFL games on Sunday, 2009 employees: 1,100 18.4 Monday or Thursday nights are considered fans, Founded: 1920 8.5 Number of teams: 32 those who participate in the NFL-inspired fantasy Viewership: 205 million football leagues might better be deemed fanatics. The 2010 people tuned in to NFL fantasy teams that individuals put together might be 20.0 8.2 games in 2013 imaginary, but the process of managing a franchise is www.NFL.com very real for them. “When Tom Brady throws that 2011 Source: National Football League touchdown pass, it’s six points for the New England 19.8 8.1 Patriots,” says Daniel Shlossman, the NFL’s director of product for fantasy football and mobile games. “But if 2012 I’m playing Tom Brady, it’s also six points for my 19.3 fantasy team.” 7.6

2013 This year, there will be 41 million fantasy players 20.3 across all sports, nearly 70 percent of whom say that 7.0

football is their favorite, according to the Fantasy Sources: NFL and The Nielsen Company. http://goo.gl/nGSRhN Sports Trade Association (FSTA), with their ranks increasing at least 10 percent every year. Around So when the NFL was looking for new ways to use three-quarters of fantasy football participants use four data to maximize the engagement of football fans, the or more Web sites for research. They spend nearly 18 fantasy market was a logical target. In addition to the hours a week consuming sports content and more direct revenues that fantasy league participation than 8.5 hours a week specifically on their fantasy generates, fantasy league members view about seven teams, FSTA says (see Figure 2, “Fantasy Sports times more content, including text, video and data, on Users Are Active Fans”). NFL.com than nonmembers. The NFL has its own fantasy sportswriters online, and NFL.com is chock full of key breaking news, expert analysis and weekly rankings to help fantasy managers develop their teams.

The NFL’s Player There has been no shortage of companies looking to FIGURE 2 Fantasy Sports Users Are Comparison Tool be the preferred destination for fantasy footballers. Active Fans Challenges The NFL launched its own fantasy football league on } Increase direct • There are 36.6 million fantasy users across all NFL.com in 2010 and has sprinted into a position sports. engagement with among the leaders. fantasy football users • 69% of fantasy users say football is their favorite fantasy sport. by developing a player comparison tool for In 2011, the league introduced in-game video, • The universe of fantasy users is growing at a rate of 10% to 12% annually. NFL.com delivering big-play highlights to viewers only seconds } Create a simple user after they happen on the field. In 2012, it added more • 74% of fantasy users go to four or more Web sites to conduct research. interface that enables functionality, such as social tools and basic research users to easily make • Fantasy users dedicate 17.89 hours a week to tools, along with simpler fantasy football games to lure watching sports. better decisions about in the novice or casual user. The league also their fantasy teams • Fantasy users spent 8.67 hours a week specifically on fantasy sports. } Enable customization so incentivizes participation in its own fantasy leagues by

that the tool is equally awarding prizes for league champions, such as trips Sources: Fantasy Sports Trade Association valuable to novices and to the Super Bowl (see Figure 3, “The ROI of experts Customer Engagement”). “We are leveraging a custom cloud platform to answer } Ensure that the online very fantasy football-centric questions,” Shlossman says. tool is scalable and Last year, as new, more advanced tools for data “‘Which player should I start? Who’s going to perform reliable, with 100 percent uptime analysis and visualization became available, the NFL better in a certain environment?’ Fans are wanting an saw an opportunity to further fuel its captive fantasy answer very quickly that they can understand.” Elements of the contingent. “In 2013, we wanted to focus on the NFL’s Player data,” Shlossman says. “We wanted to make it The tool uses 10 years of historical information to make Comparison Tool Solution simpler to understand the data, make it easier to play predictions about player performance so that fantasy fantasy football and grow that market.” franchise owners can make data-driven decisions } Deploy the tool on an in-memory analytics about their teams. They can analyze and track more platform in the cloud to Putting Big Data in Customers’ Hands than 100 stats on each player, from team defense ensure real-time As is the case with any statistics-generating sports tendencies to home-and-away performance processing, scalability organization, the NFL has for years been collecting comparisons. The tool also incorporates variables and reliability and analyzing what is now commonly known as big ranging from the forecast weather at kick-off time to } Utilize self-service data data. But that activity has taken on a sharper focus playing surface to injury reports. Historical and real-time visualization software to deliver player analysis in in recent years. “We are able to use our own data from live games is fed into a statistical model that a user-friendly format exclusive content, data and statistics to increase predicts player performance. } Enable users of the tool the fantasy market,” Shlossman says. The league to customize player wants fantasy players to compete on NFL.com. “There are lots of things we can do with the data to break analysis along five But, he adds, “at the end of the day, we want fans it down at a very detailed level,” Shlossman says. “It’s all dimensions: playing fantasy football because it brings them about the live data and understanding who the players performance, matchup, consistency, upside and closer to the game.” are and who the best players are for me to play in my intangibles fantasy team.” The question was, “how do you bring a tremendous volume of data and statistics to fans in a very A Tool for Rookies and Veterans simplified, intuitive and highly visual way?” says Vishal The behind-the-scenes number-crunching that delivers Shah, the NFL’s vice president for digital media player comparisons is very complex, but the user business development. The NFL used the same interface had to be simple, Shah says, and the tool had in-memory analytics platform and data visualization to work for the fantasy rookie and the fantasy pro. The software that businesses deploy to gain a competitive NFL wanted a tool that was just as valuable for someone edge in their industries for its Player Comparison Tool. who has been fielding fantasy teams for 15 years as it

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

“At the end was to someone who has never done it before. “A FIGURE 3 The ROI of Customer tool that works for both markets was the goal,” Engagement of the day, Shlossman says. we want • A fully engaged customer represents an average 23% premium in terms of share of Throughout the week, a fantasy football user typically fans playing wallet, profitability, revenue and relationship growth makes three types of decisions: whom to start, whom compared with the average customer. fantasy to add or drop, and whom to trade. The tool can • An actively disengaged customer represents guide users through all three types of scenarios, football a 13% discount in terms of share of wallet, enabling them, say, to evaluate a rival user’s proposed because it profitability, revenue and relationship growth trade or whether to start a particular quarterback. compared with the average customer. brings them Source: Gallup. http://goo.gl/mdjf2A closer to the Novice users can keep it simple and do a basic game.” head-to-head comparison, say, between starting the Houston Texans’ Andre Johnson or the Denver Stability and Scale on Any Given Sunday

—DANIEL Broncos’ Wes Welker at receiver. The two are very For any fan-facing system, the NFL requires 100 percent SHLOSSMAN, different players. Welker is consistent, catching short uptime. That’s quite a challenge for a pursuit that DIRECTOR OF passes over the middle of the field and delivering involves millions of users hitting the system at the same PRODUCT FOR FANTASY FOOTBALL steady fantasy points. Johnson is known for streaking time—just before kick-off. “We’re a direct-to-consumer AND MOBILE GAMES, down the edge of the field and reeling in the long, business,” Shlossman says. “[The tool must] scale to NFL rainbow passes that can end in a touchdown. The millions of users who want to access this data at any tool “kicks back a simple-to-read answer about who given moment.” is recommended in a head-to-head decision,” Shlossman says. The NFL chose to host the Player Comparison Tool in the cloud for flexibility and stability. “Everything we build has More experienced users can drill down for a more to be best in class,” Shlossman says. “We get to very, nuanced analysis of potential performance in the very high load levels on Sundays and other peak specific game—how each player performs in snow, periods—thousands of requests per second on the for example, or against a certain defensive back. platform as people make last-minute changes to their lineups. We had to build a tool that not only utilized all of Rather than rely on one-size-fits-all rankings, users this data but was also accessible and performed well at can easily customize the tool by tweaking how much the most crucial moment for fans, which is right before weight to place on five different variables: game time.” performance, matchup, consistency, upside and intangibles. If users decide they need a big week, The NFL’s fantasy apps show how consumers as well as they’ll turn up the volume on the upside and turn businesses can reap the value of real-time analytics, down the consistency dial. opening up a whole new spectrum of consumer-grade tools beyond fantasy sports. Just remember, Shah says, “Users come in and add their own spice to it,” to make the tools “immersive and fun. That’s what Shlossman says. “They can adjust the sliders if they attracts the fans and hopefully keeps them on our want big points and get one result, or tweak it and get platform for longer periods of time.”• something else.” The league’s fantasy site also features mixes of variables devised by each of the Stephanie Overby is a freelance business and NFL’s own fantasy experts, which managers can use technology writer based in Boston. at home. This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Nomura Research Institute at a glance Japan’s largest consulting and IT consulting firm

Industry: high tech

Business activities: Consulting, financial IT solutions, industrial IT solutions, IT platform services

Established: April 1, 1965

Headquarters: Tokyo, Japan

Capital: 18.6 billion yen

2012 sales: 363.8 billion yen Employees: 5,823 (NRI Group 7,738) www.nri.co.jp/

Source: Nomura Research Institute

Page 79 RESEARCH UPDATE | Real-Time Enterprise Case Studies

Tokyo Drivers Avoid Traffic Jams with NRI’s Big Data and Analytics With in-memory computing and big data behind it, Japan’s popular Zenryoku Annai! application is helping drivers get to their destinations faster and more efficiently—both day and night. BY JOE MULLICH okyo’s first taxicabs appeared a century ago. Now, about a quarter million of them service the Tmetropolitan area and beyond. For Nomura Research Institute (NRI), Japan’s largest consulting and IT consulting firm, all these vehicles crisscrossing the nation represent a huge wealth of information

and clues about how to break gridlock in one of the Nomura Research Traffic Data Jam in Japan world’s most densely populated countries. Institute at a } Challenge: 360 million data records of Japanese Glance traffic information needed to be analyzed more } Business activities: NRI is analyzing traffic jams in Japan using multiple quickly than the several minutes required by Consulting, financial IT sources, including traffic data from sensors around the traditional relational database technology. solutions, industrial IT country and location data from 12,000 taxicabs. What } Solution: In-memory computing, which can solutions, IT platform services enables NRI to keep up with the massive data flow is analyze large amounts of data quickly, was able to process the 360 million records in just over } Established: April 1, 1965 the use of in-memory computing technology, which can one second. } Headquarters: Tokyo, process reams of different data types in real time for Japan instant analysis.1 } Capital: 18.6 billion yen Japan can plot out the shortest travel routes, avoid traffic } Sales: 363.8 billion yen (fiscal 2012, ending Not too long ago, NRI relied on more traditional snarls and estimate what time they will arrive at their March 31, 2013) technologies. But the database component, in particular, destinations.4 } Employees: 5,823 (NRI became a stumbling block for this application. According Group 7,738) to Hiroshi Terada, NRI’s general manager for ERP Zenryoku Annai! combines information from satellite www.nri.co.jp/ solutions, “An ordinary relational database took several navigation systems linked to sensors at fixed locations Source: Nomura Research Institute minutes to analyze the 360 million traffic records and we along roads with traffic data determined through had problems with real-time data processing.”2 statistical analysis on position and speed information from subscribers, moving vehicles and even pedestrians.5 After NRI implemented in-memory computing technology, Meanwhile, data from thousands of taxicabs is added it was able to analyze the millions of records in just over to the mix. Using all this information, Zenryoku Annai! one second. “The speed this enables is sudden and analyzes road conditions and helps drivers plan routes significant, and has the potential to transform entire more accurately and over a wider range than is possible business models,” says Akihiko Nakamura, corporate with conventional GPS systems.

1. SAP. “NRI: HANA.” Customer senior vice president, Services and Industrial Solution Testimonial Video. http://goo.gl/gat2AC 2. SAP. “NongFu Spring: Accelerating Division, at NRI.3 “As the number of users of our Zenryoku Annai! service performance with SAP HANA.” Customer Testimonial. http://goo.gl/MAQgp1 continues to grow, so too does the amount of position 3. Ashford, Warwick. “SAP highlights customer experiences of in-memory New Business from Big Data and speed data collected. Accordingly, we must now computing at Sapphire Now.” ComputerWeekly.com, May 19, 2011. NRI is a leading example of how to find huge benefits process this data faster than ever before,” says Aritaka http://goo.gl/gsGp5e 4. Pai, Suraj. “Guest Blog: How to from “big data,” the term given to data that comes in Masuda, general manager of NRI’s Ubiqlink Department, run an intelligent business.” CIO Asia, September 14, 2012. http://goo.gl/btFQrt large volumes, variety and velocity. Thanks to big data, which is responsible for gathering the traffic information.6 5., 6. Intel. “Not limited to ERP applications alone, the Intel® Xeon® NRI has leveraged its research and traffic analysis processor E7 family also provides new business opportunities via in-memory expertise to develop a widely used service called Masuda says in-memory computing tests have confirmed technology.” Case study, 2011. http://goo.gl/6nZC9m “Zenryoku Annai!.” Using this service, subscribers all over that 360 million data points can be processed in just over

one second. The technology has improved search speed by a factor of more than 1,800 over the department’s previous The Price of Traffic Congestion in the relational database management system. United States in 2012 } Total Cost: $121 billion } Wasted Fuel: 2.9 billion gallons Besides the big boost in speed, NRI found that in- } CO 2 Emissions: 56 billion pounds memory computing helps simplify the programming needed to solve a complex analytical challenge and— Source: Texas A&M Transportation Institute15 Tokyo at a Glance somewhat counterintuitively—has even reduced the } Population: 13.2 amount of processing power needed. The NRI findings come at a time when many government million (metropolitan agencies and companies in Japan are exploring new area) “In addition to making real-time processing possible, this ways to leverage traffic data and other sources in an } Number of taxicabs: approach simplifies traffic-data generation logic,” says effort to improve quality of life issues. Toyota, which is the Approximately 250,000 Kenji Honda, an NRI senior systems analyst in the Ubiqlink largest provider of taxis in Japan, recently announced it (metropolitan area) } Number of cars: Business Planning Group. He says the higher speed of is using real-time traffic information from 700,000 Toyota Approximately 3.33 data extraction processing achievable with in-memory vehicles on the country’s roads to offer what it calls a million computing means that historical data can be processed “big data” service to local governments and businesses. } Total land area: 2,700 on a continuous basis. This, in turn, reduces the amount of The service is aimed at helping drivers during disasters square miles computer resources required to perform calculations.7 and emergencies. Drivers will be encouraged to share their own observations on road conditions, including A Better Quality of Life blocked paths and strong winds, with other drivers The ability to analyze traffic jams more quickly—and thus around the country.12 to potentially alleviate issues faster—is crucial in nation’s like Japan, which can be crippled by traffic congestion. A recent report in The Japan Times notes the Japanese During the country’s largest spring break, traffic jams can Government also wants to use car navigation data to stretch 37 miles or more.8 better monitor traffic after disasters.13 The Transport

7. Intel. “Not limited to ERP Ministry found it was able to check conditions on only 79 applications alone, the Intel® Xeon® processor E7 family also provides new Traffic congestion also can and does have a significant percent of the nation’s roadways on the night after the business opportunities via in-memory technology.” Case study, 2011. impact on quality of life and the economy in many major earthquake and tsunami ravaged the Tohoku region in http://goo.gl/6nZC9m 8. Yamanaka, Megumi and Yuji Okada. cities around the world. For example, a 2013 report from the 2011. Following the disaster, the ministry discovered that “Japan Braces for Record Traffic Jams as Aso Cuts Highway Tolls.” Bloomberg, Texas A&M Transportation Institute found that the financial traffic data collected from car navigation systems clearly April 27, 2009. http://goo.gl/9i93AX 9. Texas A&M Traffic Institute. “As Traffic cost of congestion in the United States in 2011 was $121 showed roads on which no vehicles were running. Being Jams Worsen, Commuters Allowing 9 Extra Time for Urgent Trips.” February 5, billion, translating to $818 per U.S. commuter. able to understand traffic conditions immediately, both 2013. http://goo.gl/fyNEJL 10. Hotz, Robert Lee. “The Hidden day and night, is a crucial aid in responding to a disaster. Toll of Traffic Jams.” The Wall Street Journal,” November 6, 2011. The likely negative impact that traffic jams have on public http://goo.gl/mOjIMK 11. Enterprise Innovation Editors. “Traffic health is an even greater concern. As roadways choke on All this demonstrates the growing need for—and benefits congestion top cause of employee stress, declining productivity.” Enterprise traffic, researchers suspect that the tailpipe exhaust from of—using big data in the public sector. “In the past, it Innovation,” March 14, 2011. http://goo.gl/mIDZYz cars and trucks—already implicated in heart disease and could easily have taken several hours to process large 12. The Associated Press. “Toyota to try out real-time traffic data service in Japan cancer—may also injure brains cells and synapses key to volumes of data, and for this reason, I suspect many of next month.” CTV News, May 29, 2013. 10 http://goo.gl/TNbFe5 memory. According to one analysis, drivers traveling the our clients ultimately abandoned the process of data 13. JIJI. “Government wants car navigation data to monitor post-disaster 10-worst U.S. traffic corridors annually spend an average analysis” in the public sector, says NRI’s Masashi. “But in- traffic.” The Japan Times, August 3, 2013. http://goo.gl/yRnbtU of 140 hours, or about the time spent in the office in a memory computing may well provide a way to offer faster, 14. Intel. “Not limited to ERP 14 applications alone, the Intel® Xeon® month. Meanwhile, a 2013 survey by workplace solutions tailored solutions to them.” • processor E7 family also provides new business opportunities via in-memory provider Regus suggests that traffic congestion and technology.” Case study, 2011. Joe Mullich is a freelance business and technology http://goo.gl/6nZC9m crowded public transportation systems are the top 15. Texas A&M Traffic Institute. “As writer based in Sherman Oaks, Calif. Traffic Jams Worsen, Commuters causes of stress and declining productivity among Hong Allowing Extra Time for Urgent Trips.” 11 February 5, 2013. http://goo.gl/fyNEJL Kong employees. This research project was funded by a grant from SAP.

2 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Norwegian Cruise Line at a glance Cruise ship operator that gives guests the freedom and flexibility to create their own dining and entertainment activities

Industry: travel & transportation

Headquarters: Miami, Florida

Founded: 1966

2013 revenues: $2.6 billion

Year-on-year growth: 12.9% Initial public offering: January 2013

Number of cruise ships: 13, with four on order

Passengers carried in 2013: more than 1.6 million

Destinations visited in 2013: 114

www.ncl.com

Source: Norwegian Cruise Line

Page 82 RESEARCH UPDATE | Real-Time Business Case Studies

Norwegian Cruise Line Sets Its Course for Real-Time Insights Innovative cruise company adopts predictive analytics and real-time business capabilities, with the aim of improving the guest experience and ensuring operational efficiency.

BY LAUREN GIBBONS PAUL

orwegian Cruise Line has a reputation to uphold. Founded in 1966, the N$2.6 billion, 13-ship cruise line has a track record of innovation and differentiation. The company was the first in the business to offer round-trip

cruises out of Florida’s Port of Miami. It also insights that have led to major operational cost revolutionized the industry with its signature Freestyle reductions. Over time, Norwegian’s real-time capabilities Norwegian Cruise Cruising, which gives guests the freedom and flexibility will aid in its ability to quickly ascertain and even predict Line at a Glance to create their own dining and entertainment activities. guests’ needs and desires. Industry: Hospitality/travel Headquarters: Miami No wonder, then, that an engaging and increasingly “The real power of the tool comes in predicting the Founded: 1966 personalized customer experience is paramount to future and giving visibility we never had before,” says 2013 revenues: $2.6 billion Norwegian’s quest to keep up with industry trends and Ben Lightfoot, director of business intelligence delivery Year-on-year growth: 12.9% continue attracting the growing number of people systems at Norwegian. Initial public offering: around the world who want to book a cruise in January 2013 Number of cruise ships: anticipation of enjoying a one-of-a-kind vacation (see Many companies in the hospitality/travel industry are 13, with four on order Figure 1, “Top 4 Cruise Industry Trends”). According to leveraging analytics to gain a better understanding of Passengers carried in 2013: Cruise Lines International Association, the industry will their customers’ needs and preferences. “This is what More than 1.6 million serve 21.7 million passengers worldwide, up from 21.3 the innovators are doing,” says Joshua Greenbaum, Destinations visited in 2013: million in 2013 and 13.5 million in 2009. principal at EA Consulting. “It’s a very competitive 114 environment. They are taking every piece of data they www.ncl.com Source: Norwegian Cruise Line But to maintain its track record for innovation, and have and using it to change the customer experience.” develop offers that will resonate with different segments of customers, Norwegian knew it needed to better understand who its customers are, what they value and FIGURE 1 Top 4 Cruise Industry Trends what they most want to do when onboard the ship. Improved technology to lower the cost of onboard With an eye toward increasing guest satisfaction, communications and provide more efficient passenger among other goals, Norwegian embarked in 2012 on servicing. a fast-moving journey to implement a business First-time passenger growth coming from younger intelligence (BI) system aimed at delivering insights to generation travelers, especially Millennials. business users. Because it wanted to analyze large volumes of data from multiple sources—and needed Rebound in luxury cruising, stimulated by an improving rapid results—Norwegian chose an in-memory economy and increased consumer confidence. platform, which aggregates and analyzes vast More all-inclusive options and packaging in amounts of data from multiple data sources in real accommodations, services and amenities. time. From its earliest days, the system has generated Source: Cruise Lines International Association. www.cruising.org

Voyage to Customer Insight FIGURE 2 Swimming Through the Data According to Lightfoot, the homegrown BI system Norwegian leverages its BI solution, powered by in-memory computing, for a host of data had long turnaround times and performance issues, types in real time. lacked flexibility and was too complex for anyone but Norwegian Cruise power users and IT staff to use effectively. There was Financial data. Norwegian’s BI system elucidates Line Business and authenticates the financial data in its ERP system, no true enterprisewide BI strategy in place and no Challenges something it was not able to do previously, according real processes around master data management. } Legacy BI platform to vice president of operations finance Chad Berkshire. Now, business users can swim in the detailed transac- was slow, inflexible and tion data behind the financials. complex “We knew we needed a state-of-the-art BI platform } Individual departments and strategy to stay ahead of the competition,” Inventory data. Norwegian uses analytics to keep could not easily share Lightfoot says. “The business couldn’t wait until the a closer eye on food/drink inventory aboard its cruise data, leading to multiple next day for an answer.” Further, departments ships, spotting trends and heading off waste. In the versions of the truth maintained siloed data marts, with no way to tie the future, it will use the system for inventory planning and } The company needed data together, adds Chad Berkshire, vice president of forecasting. a quicker and more operations finance at Norwegian. “There was no ‘one effective way to validate Guest data. Currently, Norwegian uses analytics to and scrutinize financial version of the truth.’” segment and better understand its guests. In the near data in real time future, it will use the system to predict which guests } Norwegian wanted In late 2012, Norwegian began its quest for a platform and prospects are most likely to go on a cruise in the to better understand near future. that offered performance, self-service for business users customer needs and and streamlined decision-making. “We knew this couldn’t Source: Norwegian Cruise Line preferences to improve and personalize the be a project that dragged itself out for the next three to customer experience five years,” Lightfoot says. “The BI platform needed to without any concerns around the data they were produce results within a year. This platform would be a consuming,” Lightfoot says. Norwegian Cruise game-changer as to how users consumed and Line Solutions performed data analytics. Changing our organizational Financial Reality Check } Adopting an analytics culture around data needed to begin right away.” The first use of the system was to provide a reality check system based on an for financial data. “We needed to deliver the transactions in-memory platform, The team selected the system in June 2013, and that made up the numbers in the ERP system,” Berkshire enabling real-time implementation progressed quickly. The ERP system says. The increased scrutiny was a natural part of analysis of a large variety was the first data source to move onto the platform, with Norwegian’s becoming a public entity in January 2013. of data types “We needed to talk to potential shareholders about } Porting ERP data the property management and reservations systems to the in-memory slated for future phases. Along with data cleansing, data specific financial goals and performance,” he says. platform, followed by governance and master data management, another property management, major change was the creation of a new corporate With the system in place, Norwegian could begin to reservations and department for BI delivery, headed by Lightfoot. obtain the coveted “360-degree view” of its guests. customer data Under the legacy data warehouse, guest data was } Implementing data maintained in multiple data marts, and understanding cleansing, data “One of the other huge contributing factors to the governance and master success of the implementation was the decision to the correlation between pre- and post-spend was data management create the BI department under the business becoming increasingly difficult. Guest information was processes umbrella,” he says. “This shift from the norm allowed sourced from the reservations system, property } Establishing a business the BI department to align itself with the business, management system, casino system and guest department dedicated to immerse itself in the business culture and drive home satisfaction. However, there was no way to bring all of BI and analytics the necessary business requirements.” It also this together without complexity. The new system is ensured system usability at all levels of the enterprise. changing that, giving everyone the same view of guests, “I wanted users to have a true self-service platform enabling them to provide a better customer experience that would allow them to swim through their data (see Figure 2, “Swimming Through the Data”).

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

Early Wins FIGURE 3 Better Decisions Through Data Soon after the system went live Berkshire was able to Business drivers for investment in analytics include the following. (% responding) see how much money was spent annually on individual Benefits of Improving the quality of decision-making In-Memory vendors across purchasing systems. “It turned out 59% Computing at there were five different departments all using one Making quicker decisions Norwegian Cruise vendor,” he says. “We had been negotiating in siloes,” 53% Line says Berkshire, whose team worked with the vendor’s } Operational cost representative to reduce the bill by 6 percent. Improving planning and forecasting reductions as high as 6 47% percent Purchasing was a fertile area for more hard returns as the Developing new products, services and in purchasing revenue streams } Expected revenue purchasing department obtained visibility across multiple 47% improvements driven systems. “We discovered that a significant number of by improved ship services we were buying were not running through a Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX configuration and new purchasing system. They were being negotiated without food and entertainment purchasing being involved,” Berkshire says. Purchasing For example, guests who meet a certain threshold of packages has reestablished oversight of services purchases, driving } Improved scrutiny and spending on the gaming systems will likely be passengers validation of financial additional cost reductions and greater efficiency. who Norwegian would encourage to book another trip. data Lightfoot expects the system will help his team pinpoint } Improved inventory The system would also potentially highlight problematic profitable customers and develop personalized offers (see planning and forecasting areas on the ships related to revenue. “We might Figure 3, “Better Decisions Through Data”). } Holistic customer views, change the way the casino floor is laid out, or place leading to improved assets in different places as a result of looking at the segmentation and “They will be able to close the gap between someone prediction of needs asset spend across the casino floor,” Lightfoot says. who is interested in taking a cruise vs. someone who is onboard enjoying themselves,” Greenbaum says. This is The biggest impact for guests so far has been the a major advantage in an industry where there are creation of onboard food and entertainment packages complex products, many customers and a lot of based on common spending patterns. “We make it competition. “There are 100 different cruises. Being able convenient for them to buy it in one package while also to take all these potential products and correlate them to gaining savings,” Berkshire says. customers based on data—that is the perfect scenario,” he says. “You can move the needle very quickly.” Another example of data-fueled innovation: Norwegian recently introduced all-inclusive cruising, the first of the Norwegian’s use of the system will increase over time, as will 1 major cruise lines to do so, according to USA Today. its associated benefits. “When we can begin to predict for With this option, available for 2015 cruises, the guest the future, the payoff will be huge,” Lightfoot says. “We are pays one price for unlimited extras, including access to now looking in the rearview mirror. The goal is to be able to specialty restaurants and alcoholic beverages, predict months in advance, so we can put together the right shoreline excursions and bingo games. offers, the right experiences for the right customers.”

Predicting the Future So far, Norwegian’s journey has been smooth sailing, with Currently, the crucial activity of customer segmentation great things on the horizon. “A lot of hard work has gone is handled by an external service provider. With the into this,” Berkshire says, “but it has really paid off.” • new solution in place, Norwegian will take that function back in-house. “In this business, you have a fixed Lauren Gibbons Paul has written extensively on customer relationship management and customer experience number of customers based on the ships and number 1. Sloan, Gene. “Norwegian Cruise management for more than 15 years. Line tests all-inclusive packages.” of rooms,” Lightfoot says. “We really have to get the USA Today, July 31, 2014. http://goo.gl/opYCzx. best guests, so segmentation is really key for us.” This research project was funded by a grant from SAP.

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

Southern California Edison at a glance Southern California Edison is a subsidiary of Edison International, which was founded in 1886. A regulated utility, SCE delivers electricity to 4.9 million customer accounts in 430+ cities and communities

Industry: utilities

Headquarters: Rosemead, California

Service territory: 50,000 square miles

Transmission and sub-transmission line miles: 12,782

Electricity wood poles: 1.3 million

2012 annual operating revenues: $11.9 billion (Edison International) Employees as of 2012: 16,593 (Edison International)

www.sce.com

Source: Southern California Edison

Page 86 RESEARCH UPDATE | Real-Time Business Case Studies

How Southern California Edison Harnesses Big Data’s Power Real-time capabilities are transforming how the electric utility manages supply and demand to lower costs and improve efficiency.

BY STEPHANIE OVERBY o keep up with electricity demand from its 14 million customers, Southern TCalifornia Edison (SCE) plans to invest more than $20.4 billion during the next four years to expand and strengthen its electric grid. But pouring money into physical Southern infrastructure such as power plants and substations is not enough. Leaders at the California Edison At a Glance regulated utility need to manage energy supply and FIGURE 1 Real-Time Data Bolsters Industry: Electric utility demand more intelligently, using more data, better Performance

Company description: analytics and faster reporting. By integrating processes, information and technology, Southern California companies deliver real-time benefits “to a very large extent.”

Edison is a subsidiary High performers Other organizations of Edison International, Many organizations, across all industries, have which was founded in similar ambitions. But for utilities, the transformation Improve the ability to analyze costs and benefits 1886. A regulated utility, to a real-time business—one that uses dynamic and 62% SCE delivers electricity granular data to make decisions in the moment—is 7% to 4.9 million customer in many ways an industry imperative. Theodore F. accounts in 430+ cities Embed real-time decision-making tools Craver, Jr., chairman and CEO of SCE’s parent and communities. 46% company, Edison International, has stated that Headquarters: 2% Rosemead, Calif. preparing for transformative change in the industry Develop and capitalize on insights about customer behavior Service territory: is critical to creating long-term value. 50,000 square miles 46% Transmission and sub- 3% Not only is demand for electricity increasing, but transmission line miles: utilities are also under pressure from regulators and Provide access to information across devices 12,782 38% Electricity wood poles: consumers to reduce consumption and lower 6% 1.3 million costs. “The only way they can meet customer

2012 annual operating expectations and shareholder demands is if they Source: Accenture, 2013 revenues: $11.9 billion replace concrete with information,” says Narendra (Edison International) Mulani, senior managing director at Accenture Understanding Business Demands Employees as of 2012: Analytics (see Figure 1, “Real-Time Bolsters 16,593 (Edison SCE’s business data, like that of most large International) Business Performance”). companies, is scattered among various systems and www.sce.com databases. Integrating it was time-consuming, which SCE has responded to these demands by initiating made analysis difficult. As a result, the utility was using Source: Southern California Edison a business transformation. New technologies and only a small portion of its data effectively. “If it takes a analytics tools deliver instant information about lot of time and effort to get information out of your electricity generation and consumption to the systems, it hampers your ability to make business utility’s customers, regulators and employees. decisions in a timely manner,” says Ron Grabyan, Real-time data is helping SCE manage its manager of business intelligence services at SCE. multibillion dollar assets more effectively, while it Oftentimes, by the time you get the data together, the partners with customers to meet their needs. opportunity is lost.”

Although it was obvious that SCE needed to pull FIGURE 2 Big Data Success Factors these disparate data sources together more quickly, The top five point to a critical alignment of business strategy, people and tools. (% responding) the business intelligence services group studied what type of information, and how much of it, people Identifying where big data can have the greatest impact needed, as well as how they would use it, before 81% choosing a technology platform. Having staff who can extract value from big data initiatives Benefits of Real- 72% Time Business at Business users wanted to do deeper analysis of Finding the best tools or resources to meet goals Southern California customer data, as well as predictive modeling. SCE 70% Edison has access to an increasing volume and variety of } Integrated data: Ensuring needed storage, compute and networking resources internal and external data—from customer and smart Business users can are in place aggregate internal and meter data to weather and traffic feeds and electricity 68% external data sources to market information—that can streamline operations Getting IT and business agreement on where big data can unearth fresh insights. and improve decision-making. By aggregating and have the greatest impact } New services: A new analyzing such information in real time, the utility can 64% tool built on the begin to transform processes across its business. So in-memory platform Source: IDG Enterprise Big Data Study, 2014. http://goo.gl/BNImWX SCE invested in an in-memory platform for its helps customers manage their electricity business intelligence systems. usage and their costs. The smart meter data became the foundation of an } Faster processes: “You have all of these different databases, and you online tool that uses the in-memory platform to show Real-time data means need information from all of them,” Grabyan notes. customers what their bills might look like under various finance professionals “You’re limited by the speed of the slowest one.” An usage scenarios. can, in effect, partially in-memory platform enables users to load extremely close the books daily. } Future use cases: large volumes of data from multiple sources into one SCE had to deliver the results, which are based on the In-memory analytics database and answer questions almost instantly. previous year’s consumption, almost instantly, or holds the potential to customers would abandon the exercise. “The customer improve response times Choosing High-Value Targets expectation is not set by the interaction they have with when power outages The platform for real-time analytics is designed to any other utility, it’s set by the interaction they have with occur and to help benefit many business functions. But successful retailers or with online companies,” Mulani says. “They analysts create new statistical models with companies choose targeted projects to demonstrate expect that same transparency and ability to interact in richer results. its value, Mulani says. According to a recent survey real time.” by IDG Enterprise Research, 81 percent of respondents say that identifying the business areas Using legacy systems, the data analysis took 40 and processes where the real-time collection and seconds. Employing the new in-memory platform and analysis of big data can have the greatest impact is analytics technologies, customers get the analysis in five critical or very important to the success of such seconds. “With increased insight, more options and initiatives1 (see Figure 2, “Top 5 Success Factors for greater control over their energy usage, consumers will Big Data Initiatives”). help us engineer a better energy future,” stated Doug Kim, SCE’s director of advanced technology, in a recent With the launch of its smart meter program in 2008, press release. If customers use energy more efficiently, SCE enabled consumers and businesses to reduce SCE does not have to build as many power plants. their use of electricity during peak periods in Everyone’s bills go down. exchange for preferred pricing. The utility’s business leaders decided that if they could get detailed Transforming Work 1. IDG Enterprise. “Big Data: Growing Trends and Emerging information about electricity usage to consumers Real-time transformation, however, involves more Opportunities.” 2014. http://goo.gl/BKeOYy. faster, they could better moderate their energy use. than deploying new technology or delivering a cool

2 Bloomberg Businessweek Research Services RESEARCH UPDATE | Real-Time Business Case Studies

new tool. “The business process has to be FIGURE 3 People Before Technology changed,” Grabyan observes. “And you have to Culture change is among the top barriers to building the analytical capabilities needed to become a real- have the appetite to make that business process time enterprise. (% citing it as a barrier) change.” Although SCE’s executive team was Resources hungry for the transformation, others were less 55% eager to see their work change. Indeed, culture Southern change is among the top barriers to building the Culture 49% California Edison’s analytical capabilities needed to become a Challenges real-time enterprise2 (see Figure 3, “People Talent/skills } Centralize data for better Before Technology”). 30% and faster customer Leadership service and predictive The key is to make it clear to people how they will 27% analytics. } Deliver instant analysis of personally benefit from new systems and Data smart-meter data to help processes. The business intelligence services team 26% customers manage their showed how real-time analytics and process Technology consumption. changes could enable employees to more easily 21% } Better manage capital access information while on the phone with a investments. Source: American Management Association, 2013 customer, cut the development time for new Southern California projects in half or close the books faster. Finance Edison’s Solutions professionals do not have to wait until the last traffic patterns are and what it is going to take to get } Deploy business week of the month to perform all financial closing- everyone where they need to go. intelligence systems with related activities. They can spread that work an in-memory database throughout the month—in effect, partially closing There are, as well, opportunities to advance the to speed information the books on a daily basis. Not every piece of predictive analytics the utility has been doing for years, delivery to customers, business users and business information needs to be delivered not only to forecast outages, analyze usage patterns regulators. instantly. But as hundreds of reports run faster, a and costs, or model customer segmentation, but also } Develop flexible more efficient workforce emerges. to create new statistical models. In the past, much of customer data and the analysis, using legacy systems, was based on predictive analytics Big Returns Take Time samples from SCE’s millions of meters and customers. models. Employee buy-in will be critical as SCE rolls out That takes time, and it has not always delivered the } Enable online access to analytics to empower real-time capabilities to additional areas. With each insight decision-makers wanted. With in-memory consumers and business potential use, there will inevitably be process change. analytics, they will be able to use the full data sets and users. That is where the real value is derived—and it takes receive richer results. time. Anyone who approaches real-time transformation with a short-term outlook is destined Ultimately, the real-time transformation will deliver new to be disappointed, Grabyan advises. “You put the ways of doing business that no one can imagine today. systems in the first year, the process changes occur “When you make changes like this, you speed up over the following years.” analytics, which speeds up processes, which speeds up decision-making, which creates exponential Take outage management. Like any utility provider, change,” Grabyan says. “It will change the way we think SCE has to deal with natural disasters, such as the about operating the company.”• 2011 windstorm that left hundreds of thousands of Stephanie Overby is a freelance business and technology customers without power. The company is looking 2. American Management writer based in Boston. Association. “Organization Culture, at putting its storm tracker program onto the Lack of Resources Impede ‘Big in-memory database to enable managers to see This research project was funded by a grant from SAP. Data.’” Oct. 3, 2013. http://goo.gl/3NUXN8. where crews are, where equipment is, what the

3 Bloomberg Businessweek Research Services Real-Time Enterprises A Research Collection

T-Mobile at a glance Provides wireless PDAs, cellular telephones, tablets as well as mobile communications, DSL Industry: telecommunications

Headquarters: Bonn, Germany Founded: 1990

Area served: Europe, United States, Puerto Rico, U.S. Virgin Islands

Employees: 36,000 worldwide www.telekom.com/home

Source: T-Mobile

Page 90 C100 M94 Y0 K47 C85 M59 Y28 K28 C65 M29 Y79 K22 C33 M32 Y72 K3 C15 M72 Y100 K26 C48 M89 Y84 K49

CASE STUDY Talent & HR Management & Finance Technology Small/Midsize Marketing Business Operations Business & Sales Using Social Media to Improve Customer Loyalty at T-Mobile

1 by the C100 M94 Y0 K47 C77 M77 Y31 K16 C54 M84 Y20 K3 C85 M59 Y28 K28 C78How M52 Y28 a K6small C73 M19 Y40 K1 C65 M29ccording Y79 K22 toC61 a M20 2013 Y78 K3 study FIGURE 1 T-Mobile: Social Media Success team within AUniversity of Massachusetts Dartmouth January 2013 T-Mobile uses Center for Marketing Research, a stunning 77 2.8 million June 2013 percent of the most successful U.S. the simple act 3.7 million companies use Twitter to communicate with December 2013 of listening to customers, while 70 percent use Facebook. 5.1 million C41 M44 Y77 K15 C33 M32 Y72 K3 C15 M72 Y100 K26 C0 M60 Y100turn K17 C28 unhappy M88 Y70 K19 C48 M89 Y84 K49 C61 M66 Y61 K52 They use other social media tools in varying Source: T-Mobile customers The company’s social media efforts have resonated with a degrees. However, frequent usage doesn’t into happy growing number of its customers. necessarily lead to expected business results. customers— T-Mobile today focuses on social media as a That’s because, according to experts, social and happy competitive advantage, using it to listen and respond media success requires a new way of to customers’ desires and concerns. It has to. With customers thinking about customers. mobile phone penetration at 78 percent of the U.S. into recom- population in 2014 and leveling off,3 T-Mobile must mendation “Companies are tuned toward delivering messages, show how it is better to grow. Social media is one of engines. but that doesn’t work in social media channels,” the key prongs in its strategy. says social media consultant Paul Gillin. “Customers don’t go to social media sites to get advertising A ROCKY START BY HOWARD BALDWIN messages.” The more a company tries to sell via “When we first started doing social media in 2012, we social media, the more likely the very people it’s tripped all over ourselves,” recalls Michelle Mattson, targeting will stop listening, he warns. “Get over the senior manager, social customer support. “We didn’t idea that you’re there to talk. You’re there to help. have a business case for how to handle social properly.” The more people you help, the more you get known as someone who is a pleasure to do business with.” The original team of 24 people—dubbed T-Force— didn’t have a clear goal either, other than to try and 1. Barnes, Nora Ganim, Ava M. That is, in essence, what wireless mobile carrier reduce some negative comments on T-Mobile’s Lescault and Stephanie Wright. “2013 Fortune 500 Are Bullish T-Mobile is doing, with significant results. The Facebook page and to explore the possible use of on Social Media.” University of Massachusetts Dartmouth number of its social media fans almost doubled analytics. “We took big risks and learned our lessons Center for Marketing Research. between the beginning of 2013 and the end (see from them,” Mattson remembers. “That education http://goo.gl/tzXPgg. 2. T-Mobile. “Awards and Figure 1, “T-Mobile: Social Media Success”). got us to where we are today, but those first six Recognition.” http://goo.gl/CWPfWp. Socialbakers, the social media analytics site, months were painful.” 3. Statista. “Mobile phone penetration in the U.S. from consistently ranks T-Mobile as a “top 10 socially 2010 to 2016.” devoted” company.2 The results are even more Among those risks: When they first started tracking http://goo.gl/ZurWud. impressive given that the program has been in complaints, the system for customer sentiment existence for only two years. tracking was manual and time consuming. SEPTEMBER 2014 Eventually T-Mobile started tracking customer © Copyright 2014. sentiment in a more automated way, using a Forbes Insights. cloud-based system that integrated with its All rights reserved. FORBES INSIGHTS 1 FORBES INSIGHTS CASE STUDY

customer relationship management FIGURE 2 Upon Further Review system. But even that process was Step 1 Step 2 Step 3 Step 4 difficult. As a result, Mattson’s team Initial T-Mobile Social media T-Mobile Social media action response reaction response tried several other products and approaches before settling on the T-Mobile 7% positive T-Force quickly 33% positive announces 37% negative communicates 5% negative cloud-based services it is using today elimination of negative feedback T-Mobile today to track everything from general Advantage to management; customer sentiment to reactions to discount program CEO promises to focuses on social rethink decision new promotions and campaigns. media as a T-Mobile 5% positive After T-Force 20% positive announces 20% negative notifies T-Mobile 8% negative competitive Mattson soon determined that her campaign to executives, CEO advantage, using group needed to articulate a clear promote switching tweets that the it to listen and strategy. Out of those discussions from Blackberry campaign will be tweaked respond to came a simple mantra: “We’re listening, we’re engaging customers Source: T-Mobile customers’ Most announcements will trigger a negative reaction of 1 to 3 percent. These two desires and where they want to engage, and we’re went far beyond that, but were saved by quick responses from T-Mobile. resolving issues.” Tactically, Mattson concerns. broke the goals down to four steps: customer used the “comment” capability to say • Identify trending issues and monitor customer “Can’t make or receive calls!” Within 32 minutes, sentiment. one of Mattson’s team had posted a URL for the • Allow the tough conversations to take place. customer to send T-Force a message directly. The • Offer quick responses in real time. formerly disgruntled customer was one of 38,000 • Make positive connections that deliver on the users who “liked” the response. brand promise. GOING BEYOND THE BASICS The sheer action of monitoring customer sentiment Most telecommunications companies take is an eye opener. “Typical negative reaction on every advantage of social media to some degree. “They campaign is 1 to 3 percent,” she says. But in two use applications like Facebook and Twitter to cases where T-Mobile launched a campaign or communicate with consumers at a rudimentary announced a service change to a popular offering, level,” says Bas den Braber, principal consultant at the negative response was off the chart (see Figure Capgemini Consulting, who focuses on digital 2, “Upon Further Review”). In the first instance, it customer experience in the telecom segment. “But planned to discontinue what turned out to be a very telcos that take social media seriously are the ones popular discount program called Advantage. In the who are making progress. They’re using social second, T-Mobile tried promoting a program to get media for marketing, to derive insights and as a customers to switch from Blackberry phones. In both real-time interaction channel to enhance brand cases, the T-Force team was able to engage engagement. They’re deploying predictive analytics executive support to respond to social sentiment, platforms to help make data-driven decisions sometimes in as little as 24 hours. across the organization. They’re integrating social media with the customer service function.” Dealing with negative sentiment is part of the job, Mattson insists, noting that some of T-Mobile’s T-Mobile’s efforts put the carrier into the category of competitors simply delete negative postings. “We leading practitioners of social media in its industry. engage with those people, and in most cases, we “We have sophisticated reporting that tells us what can turn it around,” she says. In one instance, our customers are saying about us and whether T-Force posted a message on Facebook, and a they’re happy or not,” Mattson says.

2 FORBES INSIGHTS FORBES INSIGHTS CASE STUDY

For social media monitoring, the T-Force team is FIGURE 3 Chat Strategy Goes Social using a listening tool that identifies when T-Mobile is Number of issues solved via chat Up 12% mentioned and highlights potentially negative Wait times Down 3 minutes hashtags and keywords (for example, “hate” and “never”). Sometimes the team’s efforts are made Overall handling time Down 1 minute more difficult when someone with a lot of followers Source: T-Mobile “It gives us a When the T-Force social media team applied its methodologies complains, and the complaints are retweeted. That to customer service chat, management was impressed with sense of pride to happened once with a DJ who was upset about the results. work with being dunned for a bill he said he’d paid; it turned someone who is out later that he had not actually paid it. we wanted to be when we grew up, and without a clear vision or aspiration, you’ll just flail.” super-negative, No matter what level of status the person has, and then have however, the T-Force team dives in. “It gives us a Her T-Mobile colleague Jennifer Palmer, director of that person sense of pride to work with someone who is knowledge management and social media services come back and super-negative, and then have that person come support, recommends looking at what other companies say how great back and say how great we are. That’s a key element are doing to get a sense of how you can best engage we are. That’s a of customer service,” Mattson says. with your customers. “Look at brands you admire and key element of analyze why you like that,” Palmer says. “For instance, Mattson has been so successful with T-Force’s efforts the social content of [toilet tissue manufacturer] customer that she’s been assigned responsibility for improving Charmin is funny. Is that the way you want to engage service.” T-Mobile’s chat process. Her strategy involves giving with customers? There’s nothing worse than boring —MICHELLE the chat team the same tools the social media team content. And you have to commit to delivering it at MATTSON, SENIOR uses—setting up anticipated responses to questions regular intervals.” MANAGER, SOCIAL CUSTOMER SUPPORT, and building dashboards to track results. T-MOBILE Perhaps the most important lesson is to understand the But it also involved making some structural changes. complexity of making social media work. Just because the “Sometimes we had content to be used within chat T-Force team responds quickly doesn’t mean it is that wasn’t appropriate for offshore agents who may spontaneous. The team has spent a lot of time developing not understand the nuances of social media,” Mattson relationships with marketing, customer service, public says. The solution: creating a two-tiered chat model relations, legal and senior executives. “Before, marketing that shifted complicated support issues to an onshore would just launch campaigns, and the social team would team. The results were almost immediately positive respond to them. Now, we come up with questions and (see Figure 3, “Chat Strategy Goes Social”). issues relating to how customers might respond,” Mattson says. “We have to be methodical. The last thing The number one metric for Mattson, however, is we want is to become our own news story.” • resolution. “That’s the most important thing to measure in the long term. We want to earn customer ABOUT FORBES INSIGHTS Forbes Insights is the strategic research and thought leadership prac- loyalty in the way we resolve issues across all tice of Forbes Media, publisher of Forbes magazine and Forbes.com, platforms.” whose combined media properties reach nearly 50 million business decision-makers worldwide on a monthly basis. LESSONS LEARNED Bruce Rogers CHIEF INSIGHTS OFFICER What has T-Mobile learned about social media that Brian McLeod other companies could emulate? The most important DIRECTOR, NORTH AMERICA lesson is setting a goal for your social media efforts. Writer: Silicon Valley-based freelancer Howard Baldwin has written “Set your goal,” Mattson says. “We didn’t know what about business and technology since 1987.

FORBES INSIGHTS 3 Real-Time Enterprises A Research Collection

For more information on SAP HANA,

visit: sap.com/hana and saphana.com