MIT TECHNOLOGY REVIEW CUSTOM Produced in partnership with

The Rise of Data Capital

“For most companies, their data is their single biggest asset. Many CEOs in the Fortune 500 don’t fully appreciate this fact.” – Andrew W. Lo, Director, MIT Laboratory for Financial Engineering

“Computing hardware used to be a capital asset, while data wasn’t thought of as an asset in the same way. Now, hardware is becoming a service people buy in real time, and the lasting asset is the data.” – Erik Brynjolfsson, Director, MIT Initiative on the Digital Economy as retailers can’t enter new markets and security. The pursuit of these Executive Summary without the necessary financing, they characteristics drives the reinvention Data is now a form of capital, on the can’t create new pricing algorithms of enterprise computing into a set of same level as financial capital in terms without the data to feed them. In services that are easier to buy and of generating new digital products nearly all industries, companies are use. Some will be delivered over the and services. This development has in a race to create unique stocks of Internet as public cloud services. implications for every company’s data capital—and ways of using it— Some corporate data centers will be competitive strategy, as well as for before their rivals outmaneuver them. reconfigured as private clouds. Both the computing architecture that Firms that have yet to see data as a raw must work together. supports it. material are at risk. New capabilities based on this new Contrary to conventional wisdom, The vast diversity of data captured architecture, such as data-driven data is not an abundant resource. and the decisions and actions that use tailoring of products and services, will Instead, it is composed of a huge that data require a new computing yield not only radical improvements variety of scarce, often unique, architecture that includes three key in operational effectiveness, but also pieces of captured information. Just characteristics: data equality, liquidity, new sources of competitive advantage.

But fraud detection is a cat-and-mouse In fact, many companies that are light Data Capital Creates game: As soon as the bad guys know on physical assets but heavy on data New Value someone is on to them, they try a assets—for instance, Airbnb, Facebook, new angle. Algorithmic detection is and Netflix—have changed the terms In 2013, an international ring of the only way to keep up, and training of competition in their respective cybercriminals attacked one of on mountains of data is the only way industries. While many incumbent the world’s largest online ticket for algorithms to learn the difference companies possess comparably large marketplaces. But they didn’t do between good behavior and bad. Since troves of data, they don’t exploit it anything as obvious as hacking in the 2013 attack, improved algorithmic nearly as well. These companies must and stealing credit card information. policing at the time of purchase has adopt a new mindset, Brynjolfsson Instead, they hijacked legitimate reduced fraud attempts at that ticket says: “They should start thinking of customer accounts using stolen log-in marketplace by 95 percent. data as an asset.” credentials. Then they purchased tickets to sporting and music events In this example, it’s easy to focus on A 2011 study conducted by using the customers’ payment the algorithm as the hero of the story. Brynjolfsson and colleagues at MIT methods, resold those tickets, and But it’s really just an engine; data is and the University of Pennsylvania pocketed the cash. the fuel that makes it run. Without supports the concept of data as a the data, the fraud-detection capability capital asset. Based on surveys of That kind of online theft is big business. cannot exist. Data is as necessary to In 2015, identity fraud affected 13.1 the marketplace as its ticket inventory, million people in the United States the people who manage the business, “Some financial services alone at a cost of $15 billion, according and the money that pays for it all. to Javelin Strategy & Research. Digital companies still don’t seem to thieves are often hard to spot because This simple idea has big implications. understand that they’re sitting they hide in plain sight, masquerading It means that data is now a kind of as legitimate customers while carrying capital, on par with financial and on a gold mine, and that if they out fraudulent transactions. human capital in creating new digital ignore it, the gold mine can just products and services. Enterprises At the time it was breached, the ticket need to pay special attention to data turn into a trash heap. They’re marketplace had nearly 40 million capital, because it’s the source of literally throwing away pearls accounts. How did the company figure much of the added value in the world out which ones were compromised? Its economy. “More and more important of wisdom because nobody is data scientists analyzed eight years of assets in the economy are composed looking at the data.” transaction and customer histories to of bits instead of atoms,” notes Erik identify the anomalies that indicated – Andrew W. Lo, Director, MIT Laboratory Brynjolfsson, director of the MIT for Financial Engineering this scam. Initiative on the Digital Economy.

2 MIT TECHNOLOGY REVIEW CUSTOM + ORACLE nearly 180 large public companies, recorded by social media; purchases, researchers concluded that businesses “Common observations about returns, and reorders held in enterprise that emphasize “data-driven decision the abundance of data are transactional systems. “The challenge making” (DDD) performed highest is embracing this diversity and in terms of output and productivity— misleading. Instead, the issue figuring out how to use it at scale,” says typically “5 to 6 percent higher than is one of variety—the fact that Paul Sonderegger, Oracle’s big data what would be expected, given their strategist. In addition, Sonderegger other investments and information there are enormous amounts says, data capital requires new technology usage,” said the report. of scarce, or even unique, computing infrastructure and deep “Collectively, our results suggest that understanding of how to create DDD capabilities can be modeled as pieces of data.” applications that analyze and use the intangible assets which are valued by – Paul Sonderegger, Big Data Strategist, Oracle information. investors and which increase output and profitability.” Data’s Economic In fact,“for most companies, their data industries have yet to focus on data as is their single biggest asset,” notes an asset. Identity financial economist Andrew W. Lo, For example, Lo says, “some financial To call data a kind of capital isn’t who is Charles E. and Susan T. Harris services companies still don’t seem to metaphorical. It’s literal. In economics, Professor of Finance at the MIT Sloan understand that they’re sitting on a capital is a produced good, as opposed School of Management and director gold mine, and that if they ignore it, to a natural resource, that is necessary of the MIT Laboratory for Financial the gold mine can just turn into a trash for the production of another good or Engineering. But, he notes: “Many heap.” Specifically, some financial- service. Data capital is the recorded CEOs in the Fortune 500 don’t fully services institutions gather, but don’t information necessary to produce a appreciate this fact.” Perceptions about save, valuable demographic data about good or service. And it can have long- the value of data vary widely from their clients and their activities, Lo term value just as physical assets, such industry to industry, says Lo, who is says: “They’re literally throwing away as buildings and equipment, do. “With also a principal investigator at the pearls of wisdom because nobody is data capital, if you know something MIT Computer Science and Artificial looking at the data, and because it’s about your customer or production Intelligence Laboratory. “Uber, taking up space.” process, it might be something that Amazon, eBay—these companies really yields value over the years,” says understand predictive analytics. Big- Data capital encompasses all digital Brynjolfsson, who is also the Schussel box retail stores also understand the data: truck movements captured Family Professor at the MIT Sloan value of their data.” But many other by GPS trackers; “likes” and shares School of Management.

Early Attempts to Value Data Capital

Percentage of the market value of S&P 500 84% companies that comes from intangible assets, including data and software

$8 Possible value of intangible assets, including data, trillion in the United States Sources: Ocean Tomo LLC, 2015 Intangible Asset Market Value Study; The Study; Value Market Asset Intangible 2015 LLC, Ocean Tomo Sources: Journal Street Wall

MIT TECHNOLOGY REVIEW CUSTOM + ORACLE 3 Brynjolfsson acknowledges that the substitutable. For instance, you can of competition in established markets, concept of an intangible asset can be substitute one barrel for another. But threatening incumbent market leaders challenging to grasp. “You can’t see a given piece of data, such as a price, many times larger. data the way you can see buildings, and can’t be substituted for another, such people are inevitably biased against as a consumer sentiment score, because In some cases, the changes reverberate things that they can’t see,” he says. “It’s each carries different information. “As well beyond the original markets. a blind spot. But this is something a result, common observations about For example, Lo says, “It may not that is more and more important to the abundance of data are misleading,” be obvious to a bricks-and-mortar the world economy. It’s not visible, but Sonderegger says. “Instead, the issue is businessman what Uber actually offers, it’s still something that smart managers one of variety—the fact that there are but there’s little doubt that it has had have to keep an eye on.” enormous numbers of scarce, or even a massive impact on transportation, unique, pieces of data.” fuel economy, and people’s lives.” But However, data capital plays by its when looking at data, Lo says, many own rules. “It shares characteristics • An experience good. Experience CFOs don’t focus on the business and with several other kinds of capital, goods, such as movies or books, are economic implications. “They tend to but combines them into a unique mix things whose value is realized only after view data as just a technology issue, not found nowhere else,” Sonderegger you’ve experienced them. With data, a strategic asset.” this means you only know its value after you’ve put it to use, Sonderegger says. New technologies can enable two kinds The explosively fast growth This is why choosing which movies or of disruptive strategies. The first, low- in digital services raises the books—or data—in which to invest end disruption, lets new entrants your time, money, and attention always compete directly with incumbents, but specter of disruption before carries with it a degree of uncertainty. at a dramatically lower price. Amazon an incumbent even knows used that strategy successfully in the If this all sounds too academic, retail-book industry; WhatsApp did it’s in trouble. And with the beware. Seeing data as capital is both so in messaging, pushing the price to increase of digitization and highly practical and urgent in real- zero. The second type of strategy, new- world business, because it reveals the market disruption, allows companies datafication in every industry, playbook of potential digital disruptors. to target non-consumption, reaching every company has exposure customers who previously lacked the time, skill, or money to access a product to data-capital disruption. Data Capital Disrupts or service. For example, Netflix’s The question is what to do original DVD-by-mail service was so Data capital is one of the most much more convenient than going to a about it. important assets of every online store that it put the leading brick-and- consumer service created in the past mortar DVD rental companies out of decade. Google, Amazon, Netflix, and business. Uber have all realized that data is more notes. Ultimately, he says, data is a than just a record of something that One key characteristic of disruptive non-rivalrous, non-fungible experience happened. Data is raw material for innovations is that while, by traditional good: creating new kinds of value, especially measures, they may not be as good as • Non-rivalrous. Capital equipment, digital services. And sometimes, these existing products, they compensate such as a truck, for example, can be digital services—whether offered on by being better in other dimensions, used by only one person at a time. their own or wrapped around physical such as price, convenience, or new Economists call this kind of resource products—disrupt incumbents and features. This advantage becomes “rivalrous.” It’s the same with financial reorganize entire industries. disruptive when the new offering not capital. You can invest a dollar in only only improves in its new dimensions The concept of disruptive innovation one opportunity at a time. However, of performance, but also becomes initially was made famous by Clayton data is different. “A single piece of data good enough in the old ones that the M. Christensen, a best-selling business can fuel multiple algorithms, analytics, traditional market embraces it and author and Kim B. Clark Professor of and applications simultaneously,” leaves incumbent offerings behind. Business Administration at Harvard Sonderegger says. Business School. Christensen pointed But if disruptive strategies are so • Non-fungible. True commodities, out that some technological innovations well known, then why are they still such as barrels of oil, are fungible, or allow new entrants to change the terms threatening? How can they still take

4 MIT TECHNOLOGY REVIEW CUSTOM + ORACLE incumbents by surprise? In a word, incumbent even knows it’s in trouble. speed. Some of Christensen’s original Data capital will extend the And with the increase of digitization examples, such as the steel mini- reach of algorithmic decision- and datafication in every industry, mills that marginalized integrated every company has exposure to data- mills, took decades to unfold. By making. As managers capital disruption. The question is what contrast, Uber grew to more than a incorporate more data into to do about it. $50 billion valuation in six short years. Digital services grow at blistering their thinking, more kinds of speeds because of network effects, decisions will reveal their inner Three Principles of low distribution costs, and data- driven learning curves. But their logic. Choices once thought Data Capital and How value, Brynjolfsson says, “can vanish to be strictly the domain of to Apply Them very quickly, and it can scale and be replicated very quickly as well. It human beings will become the Competitive strategy means creating unique value in a unique way, creates a situation of volatility and province of machines. winner-take-all markets.” economist Michael Porter, Bishop William Lawrence University Professor Network effects happen when each at Harvard Business School, has said. user of a solution benefits from the will use smartphones by 2020. The It’s not enough to provide products or fact that many peers have chosen the ability to immediately find and access services that your customers can only same solution. The telephone system is information online reduces consumers’ get from your company. Your company a good example, as is any social-media search costs to learn about a new also has to create those offerings in a service. Again, network effects tend service and figure out whether it’s way your rivals can’t easily copy. toward a winner-take-all outcome in a worth their time to try. given market. In his classic 1996 Harvard Business The data produced by consumers’ Review article “What Is Strategy?,” The costs for companies to distribute interactions with these services Porter described this hard-to-copy digital services are low because Internet fuels data-driven hypotheses about way of creating value as a company’s and wireless providers have invested new features to add, then generates “activity system.” Activities are the heavily to build out their networks, data about the adoption of those processes a company carries out and 2.6 billion users worldwide had capabilities. Amazon did that with every day—the way it runs marketing smartphones in their pockets at the its recommendation algorithms, and campaigns, designs products, bundles end of 2014, according to Ericsson’s Facebook did it by inserting news into offerings, provides support, manages 2015 Mobility Report. The Swedish members’ feeds. risk, and protects patents. Every technology giant’s research predicts activity uses a combination of financial, This all adds up to explosively fast that 6.1 billion people—or about 70 skill-related, technology, information, growth in digital services, raising percent of the world’s population— or process resources. the specter of disruption before an

Data’s Economic Identity

Data is: In contrast:

Non-rivalrous. A single piece of data can be used A single dollar or piece of capital equipment can by multiple algorithms, analytics, or applications at be used by only one party at a time. once.

Non-fungible. One piece of data can’t be With a true commodity, such as barrels of oil, one substituted for another, because each carries unit can be substituted for another. different information.

An experience good. The value of information The value of a durable good can only be attained can only be attained by knowing the information by possession of the physical item, not simply itself. But once known, the information can be easily knowing about it. replicated.

MIT TECHNOLOGY REVIEW CUSTOM + ORACLE 5 Because information is the only Utilities installing smart meters, resource both used and produced The rise of data capital brokerages creating mobile advisory by every activity in a company, the demands a new enterprise apps, travel sites recording all the offers digitization and datafication of more visitors don’t click on—all of these are and more daily activities has a big computing architecture. colonizing new data lands. impact on competitive strategy. The Strengthening a single tier of good news for incumbents: The tools It’s difficult to know which activities of data capital are available to all the IT stack will just create new will yield the most valuable data. The answer will vary from industry to companies, not just to startups. In fact, bottlenecks someplace else. enterprises have a distinct advantage in industry and company to company. amassing stocks of data capital because The entire stack must beef up. Naturally, a company should focus on of the volume of their interactions with activities that reinforce its competitive customers, suppliers, and partners. advantage, the things that make it The three following principles—which All activities produce information, unique. However, to make educated Lo, of MIT Sloan, describes as good but they don’t produce digital data guesses, a company should look first guidelines for data capital—show how unless they involve an application, to its biggest revenue and cost drivers, to exploit this advantage. device, or sensor. Companies that especially where it interacts with have been able to see and pursue this the outside world. Interactions with foregone data—the information rising customers, suppliers, and partners Principle #1: Data Comes off activities, places, and things like are particularly crucial because rivals are probably looking at them, from Activity so much evaporating steam—have profited greatly from it. When Google too. It’s imperative to digitize key From a data-production perspective, deployed fleets of cars onto the world’s activities before the competition does. activities are like lands waiting to be roads to capture imagery, distances, The reason: If you’re not party to an discovered. Whoever gets there first and wireless network IDs, and to activity when it happens, your chance and holds them gets their resources—in associate all that information with to capture its data is lost forever. GPS coordinates, few understood that this case, their data riches. But not all For example, the Australian the cars were amassing data capital that glitters is data gold; some activities supermarket chain Coles is that could be used to create search, are more valuable than others. experimenting with a palm-sized navigation, and ad-placement services. kitchen device for making grocery lists. Scan a barcode or just tell it you want milk, and it adds the item to an online list. Through the device, Coles can gather data not just about the items customers want, but also about how and when customers make their shopping lists, opening new possibilities for targeted ads and improved service.

Digitizing activities means involving sensors or mobile apps in the activities in some way. Datafying activities means expanding the observations you capture about them. “Datafication”—a term introduced by Kenneth Cukier and Viktor Mayer-Schönberger in Big Data: A Revolution That Will Transform How We Live, Work, and Think (Eamon Dolan/Mariner Books, 2014)—runs contrary to data- management orthodoxy, which tries to settle on the minimum dataset necessary.

6 MIT TECHNOLOGY REVIEW CUSTOM + ORACLE For instance, an airplane manufacturer card that more merchants will take, captures tens of millions of data points A data-centric approach and merchants want to take the kind from every test flight of its latest to security still incorporates of card that more people have. The passenger plane. Engineers use this same goes for application developers data to speed up the delivery of safe traditional tactics like hardening and computer users in their choices of planes, but what else will they do with the perimeter. But it also operating systems, game developers all of those observations of such a and gamers in their choice of variety of flight characteristics? Not recognizes the realities of data consoles, and potential employers and even the manufacturer knows yet. capital. Data doesn’t sit still; it job-seekers in their choice of job- But the option value on potential uses market sites. of that data is likely greater than the moves around. It gets used in In each of these examples, there is a cost to capture, store, and experiment different systems, by different with it. platform between the two markets people both inside and outside that reduces the transaction costs for the two sides. Growth on one side Principle #2: Data Tends the enterprise. attracts users on the other, and vice versa, leading to a winner-take-all or to Make More Data winner-take-most outcome. Platforms The challenge for most companies is Algorithms that drive pricing, ad increase their staying power by adding to figure out where this data-capital new features that support more linked targeting, inventory management, flywheel effect will be most powerful. or fraud detection all produce data and adjacent activities for players in While careful thinking by domain each market. about their own performance that can experts is useful on this front, real- improve their future performance. This world experimentation is the true key. Digitization and datafication of more data-capital flywheel effect creates a activities brings platform competition competitive advantage that’s very hard Businesses must develop both to the doorstep of companies that may for other organizations to overcome, infrastructure and new skills in have never seen it before: health care which makes experimenting with data experiment design, data analysis, and providers and payers, manufacturers, from new digital touch points doubly interpretation. Managing that process property developers, maintenance and important. will require new tactics to handle repair providers. For example, drones conflicting, data-driven perspectives now generate near-infrared images of For example, Uber uses a dynamic on the same topic. pricing algorithm when demand for crops that indicate nitrogen uptake rides is exceptionally high. Surge from the soil. This information can be pricing, as it’s called, raises the normal Principle #3: Platforms fed to a spreader to target the right mix prices of fares to provide an incentive and amount of fertilizer for different for more drivers to get on the road, Tend to Win areas of the field. In addition, data on bringing supply in line with demand. crop yields from seed makers and on Network effects, so important in the But the price can’t rise so high that weather-related risk from insurers growth of digital services, come in two customers leave the service in disgust. could improve output even further. types. Direct network effects—which The surge-pricing multiple, which has The maker of the tractor in the middle affect users in a single market, like risen as high as eight times the normal suddenly finds itself competing to be people on a particular social network— fare, is determined algorithmically the platform for agricultural services. are a double-edged sword: They can based on Uber’s data about supply shift into reverse as users hop to the Companies must reevaluate their and demand in different cities under next thing. Consider the collapse of strategic positioning by looking for different conditions. Because the the social network MySpace after platform threats, Sonderegger says. algorithm runs on proprietary data, Facebook’s rapid rise. “Ask where a new entrant might set only Uber can use it. More important: up as a platform between two markets Every time surge pricing runs, Uber Indirect network effects happen when in the industry,” he advises. “Then captures more data about price there are two or more markets, and consider how your company might sensitivity under yet another set of a user in one benefits from choosing digitize and datafy activities key to that conditions, fine-tuning the algorithm’s the same solution as the majority of potential platform before the would-be future performance. the other group. People who use credit disruptor gets there.” cards, for example, want to carry a

MIT TECHNOLOGY REVIEW CUSTOM + ORACLE 7 The Three Principles of Data Capital

1. Data comes from activity. Digitize and datafy key value-creating activities before rivals do.

Data tends to make more data. Data-driven algorithms create data about their own 2. performance that can improve future performance, creating a virtuous circle.

Platforms tend to win. Digitization and datafication of more daily activities bring platform 3. competition to more real-economy industries.

thinking about their intellectual- structure has been a powerful way to Data Capital property issues, thinking about data capture selected observations about and Enterprise rather than computer servers.” the real world, with broad applicability and economical use of computing Architecture The hybrid cloud environment is the resources. But this reductionist basis for data-capital computing. The rise of data capital demands a approach to creating data is now The reconfigured data management, whole new enterprise-computing being supplanted by an expansionist integration, analytics, and application architecture. Strengthening a single approach. We don’t want just selected tier of the IT stack, such as analytical observations of the world; we want the tools or data management, will just whole thing. create new bottlenecks someplace “Software is being embedded Some of these observations still do lend else. The entire computing stack themselves to relational rectangles, but must beef up. Simultaneously, the in more and more products, their schemas (the columns for each advent of cloud computing, buoyed from autos to television to row) may be different, because they by continuous price-performance capture the attributes of different kinds improvement in processing, memory, prescription medications. of customers, products, processes, or storage, and bandwidth, has kicked off They’re all generating data. transactions. This increased variety a renaissance in enterprise-computing of carefully designed structures capability. That creates opportunities to expands the diversity of data capital. Large enterprises are reinventing their analyze data, and that data In addition, the capture of more data corporate computing environments as analysis creates value.” from more activities brings data in a a new set of services delivered by cloud riot of new shapes into the stock of – Erik Brynjolfsson, Director, MIT Initiative data capital. Pictures, audio files, infrastructure. They buy some from on the Digital Economy public-cloud vendors, while they build sensor logs, geographic maps, network others in corporate data centers that relationships, and other kinds of have been converted to private clouds. capabilities of data-capital computing information all create data that doesn’t adhere to three critical principles: data lend itself to rows and columns. That new computing environment equality, liquidity, and security and figures into the data-capital equation governance. Data equality means capturing and as well, says Brynjolfsson. “Computing keeping data in its original shape hardware used to be a capital asset, and format. A new generation of data while data was not thought of as an Data Equality management software—some of it open-source, some from commercial asset in the same way,” he says. “Now, Data has a shape. For the past 40 years, vendors—departs from the relational hardware is becoming a service people the most common shape has been the model, more easily accommodating a buy in real time, and the lasting asset rectangular table of the relational greater variety of data shapes. is the data. People are increasingly database. Its familiar row-and-column

8 MIT TECHNOLOGY REVIEW CUSTOM + ORACLE For example, Hadoop, one of the most irregularly shaped datasets into new augmenting that algorithm with popular open-source projects of the arrangements to obtain previously external data requires a hybrid cloud past few years, isn’t a database at all. unobtainable perspectives on the architecture that combines public It’s a file system, like the one that stores relationships among customers, cloud capabilities with corporate data documents and spreadsheets on your products, and warranty claims, for centers reconfigured as private clouds. laptop. As such, it doesn’t require the example. Because these mashups are design of a data model before storing made without the laborious process data. Instead, developers just pour in of building predetermined models, Data Security and files, images, or database extracts as they’re easily pushed aside when a new Governance they are. question demands yet another new combination. The challenge of every chief data officer NoSQL databases, which get their is to maximize value creation from a name from using “not only SQL” to But data liquidity doesn’t just mean company’s data capital while ensuring query data, are based on a decades-old more flexibility for people. Algorithms that its use complies with applicable idea called a key-value store. Unlike benefit, too. New in-memory database policies, regulations, and laws. Security, a relational database, which requires technologies represent data in and its constant partner, governance, each record to conform to a common both a row format for writing new are necessary checks on data liquidity. set of attributes, a NoSQL database records quickly, and a column format lets each record have its own set of A data-centric approach to security still attributes. This is useful, for example, incorporates traditional tactics such when storing data from sensors in Additional abilities to visually as hardening the perimeter. But it also aircraft: Each device detects a different recognizes the realities of data capital. set of atmospheric conditions or profile, transform, and explore Data doesn’t sit still; it moves around. forces, but analyzing them together is data bring the cost of asking a It gets used in different systems, by essential. different people both inside and outside new question below the cost the enterprise. These approaches are complements, of failing to do so. not competitors. The future of data As a result, securing data-capital management is one in which diverse computing means bringing new technologies such as Hadoop data lives in the systems that are best for fast analytical queries, done and Spark, as well as new cloud suited to hold it long-term, and these simultaneously. Data-lab environments environments, under supervision. systems make the data available with with close ties to production servers Established authorization, access, all of the security, reliability, and enable algorithms with new models auditing, and encryption—for data scalability that enterprise computing and tweaked weightings to go into both at rest and in motion through the demands. Equally important, these production in hours instead of weeks, network—must extend to the entire systems must cooperate so seamlessly putting new ideas to work faster. data-capital computing environment. that analysts, data scientists, and And moving extract-transform-and- Meanwhile, during a time of ever- developers don’t need to know which load (ETL) jobs from overburdened stricter regulatory requirements, it’s data lives where. warehouse environments to Hadoop increasingly important for companies clusters makes it more cost-effective to to establish strong data-governance quickly convert data into new shapes. Data Liquidity policies. But for many companies, Achieving these improvements in that effort remains on the “to-do” However, data equality creates new list: According to recent Rand Secure problems. With a greater diversity isolation is good, but attaining them as part of a coherent architecture is Archive research, 44 percent of of data feeding a greater variety of organizations polled had no formal algorithms, analytics, and applications, better. Some of the workloads will be well suited for public clouds, such as data-governance policies; of those, 22 the need to convert data from one shape percent had no plans to implement to another skyrockets. The answer is machine-learning algorithms that identify and classify people, places, and one. (For more on securely managing data liquidity—the ability to get the and governing big data, see the Oracle data you want into the shape you need things mentioned in forum posts. white paper “Securing the Big Data with minimal time, cost, and risk. Others, such as algorithms for one- Life Cycle.”) The easiest way to see the value of to-one marketing promotions based data liquidity is in data discovery, on customer purchase history, are where analysts combine imperfect, better done behind the firewall. But

MIT TECHNOLOGY REVIEW CUSTOM + ORACLE 9 Characteristics of Enterprise Architecture Looking Ahead for Data Capital Experts predict that data’s importance will only continue to grow.

Data Equality The Internet of Things (IoT) will Capturing and keeping data in its original shape and format connect 6.4 billion devices worldwide requires diverse data management technologies working in 2016 and exceed 20 billion together seamlessly. intelligent, connected things by 2020, according to Gartner Inc. These Data Liquidity devices are generating more and more Get the data you want into the shape you need data, Brynjolfsson notes. “Software is for the task at hand with minimal time, cost, and risk. being embedded in more and more products, from autos to television to Data Security and Governance prescription medications,” he says. “They’re all generating data. That Authorization, access, encryption, and auditing—for data both at creates opportunities to analyze data, rest and in motion—must extend to the full computing environment. and that data analysis creates value.”

But that’s creating new challenges, the same way that they might shop at as well. “Having data by itself is not New Capabilities any e-commerce site. sufficient,” Brynjolfsson says. “If you from Data-Capital get the data and it just sits there, that’s Additional abilities to visually profile, not going to help you. Or if you use Computing transform, and explore data bring the it in old-style decision-making, that’s cost of asking a new question below the Data-capital computing, with its not going to help. You need to rethink cost of failing to do so. hybrid-cloud architecture, proliferation your business processes about how you make decisions.” of new data management, integration, • Data nudges. Embedded analytics in and analytic services, and collaboration applications encourage users toward Historically, managers relied on between open-source and commercial desired outcomes. These include “gut instinct” for decision-making software, takes a lot of investment to next-best actions for call-center reps, simply because they lacked the data get right. But large organizations are recommendations on e-commerce Web to do otherwise, says Brynjolfsson, taking the plunge because of the new and mobile sites, and progress metrics whose co-authored 2012 Harvard capabilities the approach delivers. in mobile apps that encourage people Business Review article “Big Data: The toward particular goals. • Data-driven tailoring. Machine- Management Revolution” explores the learning algorithms that vary their issue in more depth. “Today, it’s more • Data services. Data can also be scientific, and many managers are not output based on new input can tailor treated as a service. Some companies pricing, inventory management, and accustomed to making decisions this already create ways for applications way. It’s a whole new culture.” fraud detection. But figuring out to grab data on demand for the tasks what works under which conditions they automate. These application In addition to sharpening the power and for how long requires plenty programming interfaces, or APIs, cut of human judgment, data capital of experimentation. Data-capital the cost of using existing stocks of data will extend the reach of algorithmic computing makes these experiments capital in new ways, increasing the decision-making. As managers cost-effective to run and modify on potential value of data. incorporate more data into their the fly. thinking, more kinds of decisions will Many companies collect data that they reveal their inner logic. Choices once • Internal data marketplaces. In can share with others, after scrubbing addition to canned analytics for known thought to be strictly the domain of proprietary and personally identifiable human beings will become the province questions, data capital-intensive information. Some of that data may companies offer internal data markets of machines. This already happens with be helpful to their trading partners marketing recommendations, delivery that satisfy spot demand for new data or suppliers. Other data may prove products. Visual interfaces to Hadoop truck routing, and credit approvals. valuable to completely unrelated Areas that are still experimental data reservoirs help managers and data entities. scientists alike to browse for datasets in today, such as algorithmic analysis of cancer-screening images, will become routine tomorrow.

10 MIT TECHNOLOGY REVIEW CUSTOM + ORACLE for functions such as preventive Pre-built connectors and Oracle Big How to Start—and maintenance and asset tracking, and Data SQL make the two environments How Oracle Can Help integration into enterprise applications. work together as one unified, high- performance system so that managers Generating return from data capital Other companies have begun exploring and data scientists alike can use the is not simply a matter of adding new the potential value of their own data data they care about without having to technology to the enterprise. It’s a capital by using Oracle Big Data know which system it’s in. question of integrating that technology Discovery running on the Oracle Big with the existing enterprise architecture Data Appliance with Cloudera Hadoop Big-data integration technologies such to create sustainable competitive to create internal data marketplaces. as Oracle GoldenGate for Big Data advantage. Oracle’s comprehensive That same Hadoop cluster can also play make it easier and faster to expand the portfolio of big-data solutions, built host to Oracle R Advanced Analytics for stock of data capital in the big-data on a hybrid cloud architecture, can Hadoop, a version of the popular open- management system. And tools such help drive the competitive strategy and source statistical package modified as Oracle Data Integrator for Big Data business value described in this paper. to take advantage of enterprise-scale speed up and simplify the process of computing resources. reshaping diverse data for a variety of One of the simplest ways to begin endpoint algorithms, analytics, and experimenting with new value from Companies that have proven the applications. data is through cloud services. For value of data discovery and mash-up example, Oracle Data Cloud’s data- environments want to run big-data Bottom line: If you demand to as-a-service mixes data on more workloads with the same service- know the full value of your data than 200 million unique IDs from level agreements, security, and capital before investing in it, you’ll Oracle’s BlueKai and Datalogix with a disaster recovery as the rest of the find some of that value fading away. company’s own customer data to create enterprise environments. They So whether you’re just beginning to bring unique ad-targeting or remarketing look for enterprise-grade big-data these technologies into the business campaigns. Oracle’s IoT Cloud Service management, such as connecting the or looking to make them business provides secure connections to any Oracle Big Data Appliance to Oracle as usual, don’t hesitate to take the data-generating device, analytics Exadata running Oracle Database 12c. next step.

MIT TECHNOLOGY REVIEW CUSTOM + ORACLE 11 MIT TECHNOLOGY REVIEW CUSTOM Produced in partnership with www.oracle.com/bigdata

About MIT Technology Review Custom Built on more than 115 years of excellence in technology journalism, MIT Technology Review Custom is the arm of global media company MIT Technology Review that creates and distributes custom content. Our turnkey solutions include everything from writing, editing, and design expertise to multiple options for promotional support. Working closely with clients, our expert custom-editorial staff develops a range of high-quality, relevant content, delivering it to users when and where they want it—in digital, print, online, or in-person experiences. Everything is customized to fit clients’ content marketing goals and position them as thought leaders aligned with the authority on technology that matters. www.technologyreview.com/media Copyright © 2016, MIT Technology Review. All Rights Reserved.