EQUITY RESEARCH | Apurgiul, s2t0 21,7 2017

The US sector is Matthew J. Fassler overstored and out of step in (212) 902-6740 an era of e-commerce. But [email protected] retail is not dead; it is Goldman Sachs & Co. LLC changing. How brick-and- mortar stores employ new Heath P. Terry, CFA technologies and new (212) 357-1849 models may determine how [email protected] they survive the relentless Goldman Sachs & Co. LLC shift online. In the latest report in our Profiles in Jesse Hulsing Innovation series, we look (415) 249-7464 at the enabling technologies [email protected] emerging to leverage the Goldman Sachs & Co. LLC benefits of a physical store, whether as logistics machines optimized for Lindsay Drucker Mann, CFA distribution or as (212) 357-4993 showrooms for the next [email protected] Goldman Sachs & Co. LLC generation of customer engagement. We draw on interviews with retailers, VCs Katie Price and entrepreneurs to map (917) 343-9516 the ecosystem of enablers [email protected] and challengers bringing the Goldman Sachs & Co. LLC Store of the Future to life, and to show the forces Rachel Binder super-charging data and (212) 902-0097 marketing to understand and [email protected] influence the shopping of Goldman Sachs & Co. LLC tomorrow.

TPhReO SFItLoErS e IoN f ItNhNeO VFAuTtIuOrN e Reimagining Retail in the E-Commerce Era

Goldman Sachs does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision. For Reg AC certification and other important disclosures, see the Disclosure Appendix, or go to www.gs.com/research/hedge.html. Analysts employed by non-US affiliates are not registered/qualified as research analysts with FINRA in the U.S.

The Goldman Sachs Group, Inc. August 2, 2017 Profiles in Innovation

Contents

Portfolio manager’s summary 3 How did we get here? 3 Toward a realistic transition 4 The retailer of the future 4 The store of the future 5 …for two kinds of markets 5 Advanced data analytics: The marriage of IOT and machine learning 6 Key tools 6 Story in Numbers 11 Why we are talking about the store of the future 12 The state of retail in six charts 14 How to respond to an existential threat? 15 Bridging the BI gap 17 Disruptive retailers are forcing the issue 19 What are the stakes? 19 Who else is talking about the store of the future? 21 The role of IOT 28 The evolution of beacons use cases 29 The role of computer vision 31 The role of RFID 32 Making sense of it all; AI drive processing power 36 Defining terms 37 BI-focused applications 40 Marketing-focused applications 45 Who is investing in the store of the future? 52 What could derail the store of the future? 58 Disclosure Appendix 62

Contributing authors: Matthew J. Fassler, Jesse Hulsing, Heath P. Terry, CFA, Lindsay Drucker Mann, CFA, Katie Price, Rachel Binder, Heather Bellini, CFA, Matthew Cabral, Natasha de la Grense, Richard Edwards, Hugo Scott-Gall, Toshiya Hari, Rob Joyce, Chandni Luthra, Piyush Mubayi, James Schneider, PhD, Stephen Tanal, CFA, Alexandra Walvis, CFA.

This report continues our Profiles in Innovation series, which analyzes how emerging technologies are creating profit pools and disrupting old ones. Access the entire series below and visit our portal to see related resources.

Virtual & Drones Factory of Blockchain Precision Advanced Artificial Space Augmented the Future Farming Materials Intelligence Reality

Goldman Sachs Global Investment Research 2 August 2, 2017 Profiles in Innovation

Portfolio manager’s summary

The most dynamic story in retail today is the ongoing share shift to ecommerce.

The biggest story in retail is the fate of the 85% of retail sales currently transacted in stores. Retail is not dead; it is changing. Despite the rapid and seemingly relentless rise of ecommerce, rolling forward its recent 15% growth rate on a progressively larger base still results in 70% of retail sales via the store in five years’ time. And the fate of those sales is, in part, up to the companies that still command that market share. Those that take their fate Watch a video into their own hands and play to the acknowledged strengths of the physical store by summary of this enhancing their stores should outperform those who choose to lean on legacy business report with author practices. Matt Fassler. Go> These successful retailers will make the difficult choice of optimizing either their logistics or showroom capabilities; they will lean on vendors to feed their content and fund their displays; they will deploy services tougher to mimic online; they will of course drive digital; and they will seek out, consume, and deploy data to run more intelligent businesses.

Along this path, they will disrupt the store to enhance the store, rather than to replace it, adapting emerging technologies that bridge the information edge online retailers presently bring to the marketplace. They are merging IOT and machine learning to achieve new levels of business intelligence, beginning to use sensors to understand path to purchase, key customer triggers, and sensitivity to changes in pricing and display, and to use some of these same tools to engage in next generation customer engagement. An ecosystem of firms, some established, some brand new, is catalyzing and feeding this innovation, offering a much-needed boost to an increasingly besieged industry.

We explore innovators such as RetailNext, which provides new depths of Business Intelligence to retailers; Farfetch, which is deploying emerging technology in service of an interactive in-store experience and next-gen customer engagement; Blue Yonder, a key provider of AI services to retailers loaded with data yet starved for insights; and Go, Amazon’s emerging effort to drive a frictionless brick & mortar shopping experience, which has shocked incumbents into considering more radical in-store innovation. And we hear from experts such as futurist Steve Brown; entrepreneur Healey Cypher; and, venture capitalist Steve Sarracino, all of whom explore opportunities for innovation in retail.

How did we get here?  Retailers built large chains with real estate funded by landlords through leasing and inventory funded by vendors through generous payables terms.

 These large chain retailers were able to turn to the capital markets to fund extensive real estate build-outs, leading to the development of lifestyle centers, outlet centers, and power centers without demolishing the traditional malls these formats cannibalized. They built out category-killer businesses serving increasingly limited markets. And for decades, this strategy worked, resulting in massive wealth creation. Even today, with all the growth and promise of ecommerce and the pessimism associated with brick & mortar, 131 of Forbes’ billionaires list, or 7%, have fortunes emanating from retail. Adding in fashion, consumer goods and real estate firms designated as “fashion and retail” by Forbes raises the number to 237, 15% of the total, and ranking second behind finance & investments.

Goldman Sachs Global Investment Research 3 August 2, 2017 Profiles in Innovation

 Retailers worked to optimize their businesses, but in many instances they reached out through mass marketing and cut prices aggressively with the hope that they would capture share, settling for anonymous relationships with their customers.

 More recently, however, ecommerce has offered the extreme convenience of home delivery, at a diminishing cost, with an extreme assortment – the “endless aisle,” and, if not deep value from every retailer, price transparency that enables consumers to scrutinize price. In return, consumers by definition opt in to share their data with online retailers, with complete transparency to their behavior.

In recent years, many retailers have filed for bankruptcy while others have been closing stores. The initial reaction of the rest of the “surviving” retailers has been to mimic the mechanics of ecommerce by offering consumers the ability to order online. In many instances, firms offer the additional option of in-store pickup as well as returns and live customer service (addressing product questions, break & fix). This effort is often dilutive, as the retailer is developing additional infrastructure to service the same customer and the same order. And even if these actions are necessary, they are not sufficient for retailers to thrive.

Toward a realistic transition But there is still a long way to go between the current reality and a “Gattaca” of retail, an environment relieved of all of its inefficiencies (see the 1997 science fiction film), where groceries – the last bastion of brick & mortar – based on online’s modest share and Amazon’s planned purchase of Whole Foods – are beamed, seamlessly, to your home, unspoiled, unblemished, wholly organic and with no friction. We may be starting to order like the Jetsons – voicing our whims to Alexa, pressing a dash button on our fridge, or tapping out an order on our iPads – but most Americans still shop like the Flintstones, or at least like (and at the same store as) Lucille Ball. Amazon’s proposed acquisition of Whole Foods further confirms that the dominant US ecommerce retailer sees a future for the brick & mortar box.

The retailer of the future The retailer of the future will likely be a retailer of the past – just the most efficient version therein. A sector with excess capacity is likely to slim down at the expense of suboptimal models and to the benefit of the best positioned firms, especially in the absence of venture capital funding traditional retail.

From within the group of incumbents, the successful retailer of the future will need to operate either as an optimized logistics machine, an ultra-convenient shopping option, or an optimized showroom. Amazon, , and are highly efficient and offer no pricing umbrella to higher-cost retailers of commodity goods, and the convenience of home delivery will challenge all but the simplest, closest, and easiest shopping experiences for commodities. Retailers selling commodity goods will need to migrate their supply chains toward these hyper-efficient models. At the other extreme, stores that present a brand immersively, experientially, and interactively will still be able to battle for consumers’ attention.

The retailer of the future will also have to tighten its relationship with its customer, with high-quality data, through membership, or captured through analytics. But how?

Goldman Sachs Global Investment Research 4 August 2, 2017 Profiles in Innovation

The store of the future Historically, “stores of the future” put forth by incumbents have been newer, shinier, brighter versions of prior stores, differentiated more by aesthetics than by business processes or underlying economics. They were larger and grander, but not necessarily better. And certainly not constructed to reflect the blunt realities of omnichannel retailing – the challenge of funding multiple distribution channels, and weaving them together, most likely in service of a static customer base.

Enabling the store of the future: Three applications….

Solution I: Business intelligence Retailers need data. They need to establish better visibility on who is in their stores – either aligning a customer with their actual identity, shopping history, most recent online searches, or at least tracking the behavior of an anonymous customer on their shopping experience, perhaps with some sense of gender and age, as they traverse the box. They need to understand inventory – in-stocks and responses to changes in placement. They need to understand how different pricing stances impact the consumer’s path to purchase, and conversion. And, they need to understand how labor allocation impacts the customer experience and sell-through.

Solution II: Marketing Retailers need to enhance their in-store outreach, highlighting product with information and visual cues. They need to be able to change prices economically and then market those prices to consumers. They need to be responsive to in-store inquiries and optimally create unique in-store interfaces that transcend the consumer’s ability to capture information on smartphones.

In reality, applications developed to capture business intelligence can be deployed in service of the customer experience by feeding consumers data that is useful to the consumer, just as it is to the retailer – e.g., inventory location and in-stock information. Conversely, applications designed to enhance marketing can capture information on the nature of consumer interactions, performing business intelligence in its own right.

Solution III: Cost Retailers can deploy business intelligence tools to optimize costs – namely, payroll.

…for two kinds of markets

Mass market` For retailers selling high-volume, low-value consumables to the mass market, we see a focus on scalable solutions, particularly relating to inventory visibility and customer shopping habits. Moreover, a robust, data-hungry vendor community is eager to capture incremental data on product placement, pricing, and competitive dynamics among brands.

High-value transactions For retailers selling high-value goods, particularly in upscale and fashion businesses, we see value in driving transactions through enhanced in-store marketing.

Goldman Sachs Global Investment Research 5 August 2, 2017 Profiles in Innovation

The stakes We expect $20 billion of capex in existing capacity for our covered companies alone and $40 billion for the retail sector overall; these dollars are currently allocated to some combination of supply chain, ecommerce, and IT. Even if store counts downsize, this capex pool could remain intact as dollars are shifted from store growth to store enhancement. That said, the ROI on some of the potential solutions we discuss in this report is such that significant returns are feasible with far more modest financial commitments. The venture capital community is converging around the opportunity as well (see a compendium of venture investments on our Venture Horizons section beginning on page 52, and an interview with Steve Sarracino of VC firm Activant Capital on page 56).

Advanced data analytics: The marriage of IOT and machine learning Elements of solutions now driven by IOT and AI/machine learning have been in place for years. Cameras have monitored stores in service of loss prevention, RFID technology has been used to track inventory, and firms have used sensors of various sorts to track traffic through the door. But newer technology is enabling more advanced deployment, with more granular analytics that are custom-designed and yielding superior speed to market.

Retailers can now tap technology solutions to help them enhance both their business intelligence and their marketing goals.

 Sensors are now capable of tracking activity – through photography, video, and wireless signals – and conveying this information for processing simply and inexpensively. Similarly, devices can receive and deploy insight and information – e.g., posting a new price, or a tailored marketing message.

 Through AI, and enhanced by machine learning, companies can interpret observations and deduce their implications. Massive volumes of individual observations from sensors can be interpreted to yield commercial recommendations.  This volume of computing is enabled by the cloud.

Key tools  Smart shelves, that track planogram (i.e. store/shelf map) compliance, monitor inventory integrity, and provide real-time feedback to the associate in the aisles to fix the problem.

 Smart floors, monitor foot traffic, analyze consumer behavior and report back with actionable insights.

 Audio analytics, to assess traffic levels.

 Video analytics, to classify consumers into demographic groups, by age, gender, ethnicity, and then track and understand their behavior, aided by emotion recognition software, layering in notions of customer sentiment and engagement to a video-driven dataset.

 Staffing software that tracks feeds on shopper counts, dwell time, frequency of visits, and conversion rate, and deploys this data against staffing insights to optimize labor scheduling.

 Smart pricing software that address retailers’ shortfall in providing dynamic pricing, a hallmark of Amazon’s competitive offer.

Goldman Sachs Global Investment Research 6 August 2, 2017 Profiles in Innovation

 Smart dressing rooms that engage “smart mirrors,” using RFID technology to identify an item brought into a dressing room, aiding retailers’ knowledge of what makes its way into the fitting room and hence close to purchase, and enabling interaction with the consumer on the cusp of a transaction. For more color on this approach, see our interview with Healey Cypher, founder of Oak Ventures, on page 48).

Exhibit 1: A new ecosystem to drive business intelligence, innovative marketing, and sales

Inputs Sensors External inputs

• Beacons • RFID • Weather • Bluetooth • Video/photo • Calendar • Wifi • Audio • Competitor pricing/promotions • Computer vision

Processing / Observations Business intelligence tracking AI/Machine learning

• Emotion recognition • Path to purchase • Traffic counts • In-stock • Planogram compliance

Selling tools Consumer-facing applications

• Smart mirrors • Smart floors • Dynamic pricing signage

Outcomes Commercial outcomes Business process outcomes

Sales • Product placement • Pricing • Staffing • Checkout • Clienteling • Inventory optimization

Source: Goldman Sachs Global Investment Research.

An emerging ecosystem of companies is providing these solutions (see, for example, our vignette on RetailNext on page 44), while an array of legacy leaders and newer firms is enabling the underlying technology (see our ecosystem, illustrated on page 10).

The most extreme expression of this new opportunity set is Amazon Go (see our vignette on page 50), a one-of-a-kind store in beta mode, open only to employees so far, that enables consumers to walk in, check in, grab goods, and walk out with no check-out process. Skeptics might seize upon these innovations by saying, “a store like this is the store of the future – and always will be,” as it might seem aspirational and unlikely to function in the real world. But it contains elements we expect its brick & mortar competitors to pursue aggressively. Farfetch, a British firm providing an online platform for luxury brands, is developing the contours of a store that deploys technology to enhance the consumers’ in-store experience and drive conversion through its “augmented retail” approach (see our vignette on page 47).

Goldman Sachs Global Investment Research 7 August 2, 2017 Profiles in Innovation

Barriers to innovation and implementation  Improvement in ecommerce technology could replicate or replace some of the most critical functions served by a physical store. Rapid volume declines could quickly undermine the economics of store-based retailers, rendering the notion of investment moot. Also, retailers could choose to focus on their online-only offerings. The cost of emerging tools is high, and retailers may not be in a position to invest in them – and waiting for cost curves to fall may lead to resources fading faster than the cost of intended investments. Advancements in voice-based or “conversational” commerce or AR/VAR could enhance the at-home shopping experience at the expense of the store. Finally, the speed of innovation – online and in-store – could outpace the pace of execution on current investments.

 We also see potential for a lack of sponsorship within retail organizations – who takes ownership for this high-risk/high-reward investment?

 Once a management team is strategically committed, execution can prove choppy, based on technology and on institutional buy-in.

 Lastly, regulations around privacy are inconsistent and could prevent retailers from maximizing the potential impact of the available technology.

Goldman Sachs Global Investment Research 8 August 2, 2017 Profiles in Innovation

This page is intentionally left blank.

Goldman Sachs Global Investment Research 9 Machine Learning / Data Analytics Emotion / Facial Recognition Alteryx Affectiva Altierre eyeQ Amazon Hitachi Apple Panasonic Blue Yonder Reflektion Celect Sharp Consulting firms Facebook Google Hortonworks IBM Intel Robotics RFID Clienteling Microsoft MicroStrategy Bossa Nova Alien Technology MadMobile Nvidia Daifuku Avery Denison Salesfloor Oracle Fellow Robots Detego Tulip Retail Profitect Simbe Robotics Impinj Revionics Starship Technologies NXPI Semiconductor SAP Teradyne Tyco Retail Solutions Tableau Zebra Technologies

Business Intelligence Employee Management Predictspring (Beacons / Sensor Technology / AI) Theatro Beabloo Zinc Estimote Euclid Analytics Hikvision InMarket Smart Shelves / Pricing Master Card (Applied Predictive Technologies) Altierre Meridian / Aruba Displaydata Netgear (Placemeter) Impax Media Percolata Oak Ventures PlaceIQ Perch Interactive Prism Skylabs Omnichannel Powershelf / Compass Marketing Trax Qualcomm IBM RetailNext NewStore Scanalytics Reflektion Swirl Salesforce Wireless Connectivity Unacast Wynd Videomining Broadcom Cisco Qorvo Qualcomm Skyworks Zebra Technologies ROOM TO RETROFIT FUNDING GAP

Our estimate of total capex for the Venture funding in e-commerce retail sector. Even if store counts and digital commerce enablement decline, this capex pool could remain in 2016 vs. only $324mn invested $40bn intact as dollars are shifted from $14bn in store-based retail innovations store growth to store enhancement such as RFID systems. (p. 52) via new technologies. (p. 19)

ANNUAL REWARDS NEW MARKETS $253bn per year $900mn per year Our cost estimate for outfitting our US retail coverage The annual amount US retailers lose from overstocking or with sensor-based analytics capabilities—just one of understocking merchandise—a problem that could be the technologies we see powering the Store of the abated by smart inventory technologies like RFID tags. Future. Global and private adoption would make the (p. 20) total market a multiple of this. (p. 51)

$18bn per year RISE OF THE MACHINES The estimated revenue boost to retailers from a 5pp The expected increase in AI computer vision improvement in traffic conversion (i.e., percentage of revenues for the consumer sector between shoppers who make a purchase)—far less than the 2015 and 2019, according to Statista. We improvement promised by some vendors of in-store 8x see computer vision working with other marketing technologies like interactive mirrors. (p. 20). Store of the Future technologies to provide a comprehensive view of consumer and +3.2% per year product flows. (p. 31) Potential revenue improvement retailers could attain INTEL AFOOT from smart pricing technologies like digital signage, based on a 2012 Deloitte study. (p. 42) The size of “smart mat” panels— paper-thin, pressure-sensitive floor mats that can track shopper traffic 2 x 2 ft patterns and behavior—made by LONG LINES BI tech vendor Scanalytics (p. 40)

The amount of data amassed for a The maximum range—roughly typical chain, according half a football field—for retail beacons, which have 50 20bn to AI-firm Blue Yonder—a daunting amalgamation of information that can transitioned from devices used be interpreted only through enhanced to feed information to processing capacity like artificial consumers to devices used to meters rows intelligence. (p. 36) gather information about shoppers in store. (p. 29) August 2, 2017 Profiles in Innovation

Why we are talking about the store of the future

Overbuilt. Overstored. Not the end. The store still matters. 85% of retail sales are transacted through stores today. The range of sales online varies by category, with food (<1%), pharmacy (1%), and pet food (5%) toward the low-end of the range, and consumer electronics (47%), books (34%), and nutritional supplements (17%) toward the high-end. Even if ecommerce maintains the 14.4% CAGR evident over the last three years, stores will still represent 76% of retail sales five years from now.

Moreover, there is a massive labor force deployed in service of store-based retail. Over 10% of the US labor force is employed by store-based retailers, in addition to those employed by the connected ecosystem (restaurants, theatres).

Yet the retail store as it is currently constructed is in many ways an anachronism.

 Over the past century, retailers have built boxes based on their best sense of consumer needs and tastes, their analytics on site location and local demographics, and landlords marketing of their existing sites versus these same criteria.

 Retailers were able to open new concepts with limited information and limited insight about the probability of success, as costs to open a store were relatively low, given the ability to rent property with minimal money down and to purchase inventory on credit.

 By the time a concept attempted to scale, most frequently its efficacy, at the “box level,” had been proven. Its ability to scale successfully depended on whether it accommodated different regional tastes, managed to scale operations via deployment of management of technology and capable management, and tested changes to its core box with the same scrutiny by which it tested its original prototypes.

 While the process was imperfect, by the time a concept reached the stage of a large-scale rollout, its chance of success were reasonable.

 Once in business, these firms typically maintained a mass approach, selling a centrally devised, scaled value proposition to its customer base as a whole.

 The store's value proposition was marketed through brand-building broadcast and print media, with the latter comprising both newspaper ads and more targeted direct mail, but with limited personalization.

Management tactics have not been static. Retailers have improved the efficiency of inventory management, tracking sales, orders, and markdowns to optimize working capital and margin. They have improved labor management models, benchmarking payroll hours against traffic and transactions. They have improved their site selection, tracking performance alongside an array of tenants.

Still, store-based retailers have been unprepared for the onslaught of online retail.

 Retail traffic is fading as market share moves online.

 The cost of operating redundant supply chains, most likely to cater to a static customer base, is significant, and returns on increasing consumers’ fulfillment options are most likely dilutive.

Goldman Sachs Global Investment Research 12 August 2, 2017 Profiles in Innovation

Omnichannel capabilities – “of course” The cost of competing – most often of merely existing for most retailers – is serving customers where they want to shop, how they want to shop. At a very basic level, they have not optimized direct-to-consumer supply chains, centralized distribution that would allow them to satisfy consumers’ demand to shop online efficiently. They need to offer online efforts that leverage their purchasing power, brand, and existing distribution capacity (particularly their direct-to-consumer capacity, if any); allow for in-store pickup – and returns; maintain price consistency – not enabling consumers to arbitrage its own pricing; and, marketing directly to consumers.

This story has been told many times: retailers are responding, both through standalone direct facilities and through omnichannel efforts synthesizing in-store and online ordering; their margins are falling; and many are struggling to keep pace with Amazon, or to fight through the headwinds of a failing ecosystem, where traffic is drawn by a collection of firms, the failure of the weakest among them undermines that atmosphere, and the traffic and economics of the overall environment. These issues are most acute on the mall.

But store-based retailers still operate…stores.

Goldman Sachs Global Investment Research 13 August 2, 2017 Profiles in Innovation

The state of retail in six charts

Exhibit 2: Ongoing growth in ecommerce begs retailers Exhibit 3: Ecommerce share gains accelerating, as growth to invest in the store of the future persists at a constant mid-teens pace off a steady base “Core” defined as retail sales excluding auto, food, fuel YoY change in ecommerce as percentage of “core” retail sales

ecommerce as % of "core" retail sales % yoy change in ecommerce as % of "core" retail sales

16% 1.6%

14% 1.4%

12% 1.2%

10% 1.0%

8% 0.8%

6% 0.6%

4% 0.4%

2% 0.2%

0% 0.0% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Source: Census Bureau. Source: Census Bureau.

Exhibit 4: Public retailers comping below sector growth Exhibit 5: Retail margins under pressure, despite healthy rate; apparel / mall ecosystem faring worst macro on slow sales, dilutive omnichannel investment Sales-weighted indexes Retail EBIT margin equal weighted 5.0% 13.0% Retailers fall below overall 4.0% consumption, losing share 12.0% 3.0% "Core" Retail 11.0% Retail SSS Index 2.0% Apparel 10.0% Retail EBIT margins 1.0% Dept. Stores remain under 0.0% Discount 9.0% pressure. Hardlines -1.0% 8.0% Food -2.0% Mall under pressure 7.0% -3.0%

2013 2014 2015 2016 2017E 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017E

Retail EBIT margin equal weighted

Source: Census Bureau, company data, Goldman Sachs Global Investment Research Source: Company data, Goldman Sachs Global Investment Research

Exhibit 6: Strip centers are the majority of retail footage; Exhibit 7: Retail remains a large employer, but fading lifestyle, regional, super-regional properties only 17% Sales-weighted indexes

Year % of Total Non-Farm Payrolls Store employees per sq. ft. Store-based retail Non-store retail Super Regional Other, 2% Mall, 10% 2000 11.2% 0.4% 2.5 Community, 2001 11.2% 0.4% 2.4 Regional Mall, 25% 2002 11.2% 0.3% 2.3 5% 2003 11.1% 0.3% 2.3 2004 11.1% 0.3% 2.2 2005 11.1% 0.3% 2.2 Power Center, 2006 10.9% 0.3% 2.2 13% 2007 10.9% 0.3% 2.1 2008 10.8% 0.3% 2.0 2009 10.7% 0.3% 1.9 Convenience, 2010 10.8% 0.3% 1.9 12% 2011 10.8% 0.3% 1.9 2012 10.7% 0.3% 1.9 2013 10.7% 0.3% 1.9 Lifestyle, 2% 2014 10.7% 0.4% 2.0 Neighborhood, 2015 10.6% 0.4% 2.0 31% 2016 10.6% 0.4% 2.0

Source: Costar. Source: BLS, Costar, Goldman Sachs Global Investment Research

Goldman Sachs Global Investment Research 14 August 2, 2017 Profiles in Innovation

How to respond to an existential threat? At the most basic level, we see five operating principles that will distinguish the successful retailer of the future.

I. Optimized showroom or optimized distribution. It’s tough to be both. Do we need stores?

No.

Different consumers need different services. Some need to prioritize cost, and may have more time than money. Some need to prioritize convenience, and may have more money than time. Some may want to be educated about a product or a project. Others may want to be entertained. We need stores only to the extent that they optimize for one of these outcomes, and we see the best prospects for retailers that cater to either of these extremes.

Amazon is deploying robots to pick/pack/ship, growing its distribution capacity by over 30% per annum and bringing economically viable direct-to-consumer delivery closer to more consumers across more categories each year. Competing with AMZN’s cost structure and working capital commitment is challenging even for other pure-play ecommerce firms. Retailers seeking to compete in volume businesses – with cost structures that enable competitive pricing – will have the greatest chances of success if they sell through stores that optimize the cost of fulfillment. The best examples are Costco and Home Depot, each of which has always operated within the aesthetics of a warehouse with structural advantages on the cost front. In fairness, this is only a partial explanation of Costco’s success, which also relates to its focus on a narrow array of SKUs, maximizing its buying clout and of course subsidizing its costs via member fees. And it is only a part of Home Depot’s formula, which has harnessed compelling customer service, educated consumers on projects, and marketed aggressively to pros not served as well – yet – by online options. Another solution is optimizing for convenience – retailers are competing with simple online ordering and rapid delivery, and a retailer that offers a nearby shop, selling essentials at competitive prices, can also make its way in this world. The European hard discounters and and the sector are likely to weather this era well on that basis.

Relatedly, brick & mortar retailers can create an appealing experiences in store, with fixturing, imagery, and service that are difficult to match online. These boxes are unlikely to yield optimal distribution economics, except for high-value, high- margin, highly mobile goods.

The most notable examples today are brand flagships, far more focused on marketing the story than moving goods in-store.

 Nike operates a 50,000-plus square-foot space in ’s SoHo neighborhood, with intricate, spacious displays for new product, play areas for basketball and soccer, kinetic sensors that track body motion, and touchscreens than enable customized t-shirts, all of which are intended to bind the consumer to the brand. Also, consumers are connected to the store via the Nike app, and if they choose to self-identify the company can enhance their experience upon their return.

 The RH story has been volatile, and recent results have raised question about its operating model, but seeks only to merchandise its brand, leaving fulfillment to its DTC operation.

 On the mall, L Brands has the advantage of combining category leadership, colorful displays, and appealing economics (ticket/margin/cube) that enable it to generate meaningful gross profit dollars for a modest box.

Goldman Sachs Global Investment Research 15 August 2, 2017 Profiles in Innovation

The brand messages projected here are likely restricted to major metro markets with heavy foot traffic – locals and tourists – though the imagery and familiarity of these stores is likely to project beyond these locales given the reach of social media.

Inventory-free stores that enable consumers to sample inventory, i.e., try on clothes or glasses, are also promising. Consumers de-risk size, color, and style and can have orders fulfilled directly. Bonobos (through its Guideshops) and Warby Parker have successfully deployed these formats. And what better validation of the utility of brick & mortar than stores opened by digitally native brands that achieved measures of success without them? For better or for worse, retailers with sunk investment in real estate will have a harder time redirecting their in-store presence to these innovative formats.

But many businesses are “stuck in the middle.” Target combines a fashion-forward apparel environment with low-priced basics, and carries consumables in relatively modest quantities with aesthetically appealing but costly store standards. Whole Foods, even as it carried out its mission of purveying premium food in an upmarket environment, has struggled with pricing and costs and has not been able to develop a cohesive online strategy. Department stores stock mid-price apparel out of inefficient onsite stockrooms.

The most successful stores of the future will need to pick sides.

II. Alliances with vendors to optimize content and share cost Vendors retailing goods with any technical or fashion element will always want some physical outlet to display their wares. Ecommerce/direct-to-consumer options may offer superior economics and appeal to emerging shoppers as digitally-native younger generations age, but for fashion, electronics, furniture, jewelry, and other categories where the tactile element plays a role, we expect brands to want to preserve the option of an appealing live experience. In these instances, we believe that retailers that intertwine themselves more closely with leading brands are more likely to prosper. The sporting goods retail sector has been decimated by bankruptcies, but Dick’s, the most successful firm still standing, has long leaned on vendors to fund fixtures highlighting their brands. Best Buy has staged a wondrous recovery, emerging from clear Amazon victimhood to operate a sustainable business model by aligning closely with Sony, Microsoft, Samsung, Apple and others, who invest in fixturing and in staffing to ensure high-quality display and execution.

III. Services – complementing the traditional retail offer Some of the more resilient retailers in today’s environment weave a menu of services into their offerings. Home improvement retailers mix paint and offer design and décor input. Best Buy offers extensive in-store repair services (as well as outbound technology services). Walmart and Costco integrate services onto their peripheries (optical, pharmacy) or their parking lots (gas).

IV. Driving digital Rapidly growing online businesses alongside store businesses is essential; retailers need to satisfy consumer demand however their customers want to shop. But the magic stops with the customer experience. Retailers rarely disclose the impact of channel shift on profitability – if they can discern it. We have illustrated that merely shifting consumers from one channel to the next, adding redundant distribution to satisfy the same demand – is intrinsically dilutive to financial returns. And a strong omnichannel focus does not mean that a firm is making the most of its retail assets.

A seamless connection between online customers and the store is of course highly desirable. But many of the most accomplished firms in mobile commerce do not face the precise challenges retailers are trying to solve.

Goldman Sachs Global Investment Research 16 August 2, 2017 Profiles in Innovation

Starbucks captures massive amounts of information about its customers. But SBUX is selling a limited menu, with a high service component and, for on-the-go customers, a focus on speed – there is little discovery value in incremental visits to a Starbucks store. Similarly, Domino’s restaurants primarily serve as delivery depots – its mobile prowess puts it ahead of rivals in changing the node of its interface with its customers, but not their journey through a physical store.

Apple operates incredibly productive stores and is innovating further under the new leadership of Angela Ahrendts, whose work at Burberry over the past 10 years continues to influence retail store design and the integration of digital into retail brands. Most brick & mortar retailers would be lucky to catch some of Apple’s special sauce in a bottle. But the store sells a single brand, with a limited SKU count, and is focused on embracing consumers into its services umbrella.

V. Membership and data: increasingly potent tools Retailers can bind consumers to their businesses and capture critical data through membership programs, co-branded or private-label credit cards, and simply convincing consumers to download their apps. Some combination of these efforts is essential to help firms optimize marketing and merchandising decisions. That said, we have seen limits to both consumer buy-in and to retailers’ abilities to capitalize on data when they possess it. Consumers are most likely to opt-in to member programs – paid or unpaid – from high- frequency retail partners – supermarket or warehouse club – or when the economic proposition is compelling. For instance, RH runs a low-frequency business model but compels consumers to opt-in to membership with a discount that defrays the cost on any typical furniture order. The app front is tougher; apps are free, but buy-in is limited by consumer capacity, as the average consumer has 4.2 apps on their phone, according to an October 2016 survey from Google. The average may be a bit misleading, as 13% have none, 12% have one, 19% have two, and the remaining 56% have three or more; getting to a 4.2 average means that these 56% have an average of 6.6, suggesting that for a group of mobile enthusiasts, capacity is less of a constraint. Still, even though the app on a mobile device can aid a retailer’s data capture without a consumer’s active opt-in in any given transaction, unless consumers use their app, credit card, or other affinity vehicle upon purchase, the retailer’s visibility is limited.

Bridging the BI gap But even these principles reflect legacy technology, and have been applicable for years, if not decades. The combination of the information edge offered to customers by online retailers and the convenience of home delivery suggests share loss to online will continue regardless of how retailers respond. But they can mitigate the impact by bridging the gap between their skills and those of online competition. Beyond applying legacy best practices, retailers would stand a better chance of thriving if they optimized the resources unique to those stores – labor, space, and working capital – by deploying incremental business intelligence and marketing tools to drive conversion, manage inventory, and monitor labor.

Online retailers typically know who the consumer is when they start shopping the store, based on cookies, loyalty / Prime membership, customer-to-cloud feedback (i.e., via voice recognition), or the use of an app. They track customer movement, browsing, cart behavior, and where the customer has come from / heads next. They can constantly adjust their presentation, at a low cost, and customize their marketing to customers based on what they’ve learned through their visits. Moreover, online retailers can change prices rapidly, presumably based on dynamic insights related to cost and elasticity, and can vary price by customer based on any number of variables.

Goldman Sachs Global Investment Research 17

RETAIL BUYING EXPERIENCE

To enter or not to enter?

Which way do I go? AWARENESS How do I find what I know I'm What product appeals to looking for? me as I browse?

Which item do I want to pick up, hold, try on?

I'm in the fitting room; • I need another size • I'd like to order this online right now

CONSIDERATION How is it priced?

What is located adjacent to the product?

Do I need assistance? If so, is it there?

Check-out CONVERSATION Sources: GS Research, KPMG August 2, 2017 Profiles in Innovation

Disruptive retailers are forcing the issue Innovative entrants to the brick & mortar space are forcing the issue by introducing prototypes that are radically reconceiving elements of the in-store experience.

Amazon has introduced Amazon Go, on a test-basis, open only to associates in . The store uses machine learning, computer vision, sensors and other technologies to allow customers to make purchases by logging in via a smartphone app upon store entry, select products from store shelves, then leave the store without a checkout line. The store sells ready to eat meals, staples, meal kits and other convenience items. Purchases are automatically charged to the individual's account. While this concept has not yet opened to the public, and the challenge of proving out its smaller box size suggests its learnings may not prove applicable to bigger-box operators, it is ironic – and chastening to leading store-based retailers – that the dominant online disruptor in the US is developing the most disruptive store-based technology as well.

Farfetch, a UK-based online platform that focuses on luxury apparel, jewelry, and accessories, recently developed a “store of the future” concept, showcased to the media in lab form in , that deploys technology to enhance the consumers’ in-store experience and drive conversion through its “augmented retail” approach. At the heart of the effort is a “Connected Rail,” racking that leverages (1) RFID to track product and ultrasound technology to track movement, enabling consumers who have signed in to add a product to their online wish list simply by picking it up, (2) an interactive hologram that consumers can use to peruse goods online, and mix/match combinations, and (3) smart mirrors, where consumers can capture product info, receive suggestions from salespeople, and pay for their goods. These activities provide the retailer with a feedback loop to gauge interest in key products, enabling it to adjust placement and visual merchandising in addition to serving as effective marketing tools. Farfetch has acquired UK retailer Brown’s, and will introduce the technology through its stores and through Thom Browne, in New York, where the technology will also be implemented.

Hema Xiansheng, also known as Hema Fresh, or Mr. Hippo (Hema is Chinese for Hippo), operates 13 stores in China, the majority of them in Shanghai, focusing on food and dry grocery, with a focus on fresh. The store explicitly serves multiples purposes: it is a traditional supermarket, a food court for third-party sellers, and a depot for picking/packing/shipping online orders for delivery or pickup. The business was founded by a former JD executive and has received significant funding from Alibaba. Online customers can only order via an app, and the store – online and offline – only accepts payment through Alipay, BABA’s payments system. The use of the app ensures comprehensive data capture, enabling a personalized product page. Customer retention is impressive to date and online penetration exceeds 50%.

In December 2016, Italia introduced a “grocery store of the future” in Milan in partnership with Accenture. The effort provides consumers with extensive data, with a focus on nutrition and wellness, with a goal of increasing the transparency of the shopping experience.

What are the stakes?

Big capex wallet Retail sector capital expenditures totaled $85 billion in 2015, based on Census Bureau data. We expect retailers in our US coverage will spend $40 billion in capex in 2017, up slightly from $39 billion in 2016. Surveying 14 of these retailers that offered detailed break-outs of their capital spending, representing nearly half of that global spend, an average of 34% was allocated to IT and supply chain. This suggests more than $13 billion allocated to areas that enhance both omnichannel and IT capabilities for our covered companies alone. But

Goldman Sachs Global Investment Research 19 August 2, 2017 Profiles in Innovation

many of the companies that break out capex to that level of detail are still growing stores, far more than the average retailer today, and almost certainly more than retailers will be as space consolidates in the years ahead. We expect at least 50% of capex to be reinvested in existing capacity, suggesting $20 billion for our coverage, and $40 billion for the sector overall, much of which we expect to be allocated to ensuring the evolution of the store of the future.

Exhibit 9: Total capital expenditures across retail based Exhibit 10: Of the retailers within our coverage providing on Census Bureau data reached $85bn in 2015; retail detailed capex breakouts, an average of 34% of capital companies within our coverage spent $39bn in 2016 and spending went to investments in IT and distribution are expected to surpass $40bn in 2017 Spending on IT and distribution as a % of total capex Total capex spending; $ in billions spending

IT/supply chain Total capex spending ($mn) investment as a % Retailers in Total retail of total capex our coverage industry BBY 64% 2013 $38.0 $77.5 TJX 55% 2014 $37.9 $82.4 KSS 46% 2015 $38.9 $85.8 WMT 39% 2016 $39.1 - TGT 38% 2017E $40.1 - BURL 35% ROST 30% Total retail capex based on Census AEO 29% Bureau data reached $85bn in 2015. JWN 28% Within our coverage, total retail capex TSCO 27% was $39bn in 2016 and is expected to ULTA 26% surpass $40bn in 2017. MIK 24% LB 22% JCP 16%

Average 34%

Source: Company data, Census Bureau, Goldman Sachs Global Investment Source: Company data, Goldman Sachs Global Investment Research. Research.

15 million jobs Moreover, over 15 million jobs are tied directly to store-based retailers. An accelerated reduction in footage would exacerbate the necessary and ongoing push toward efficiency that is already reducing the number of retail employees per square foot of space.

Optimizing traffic Apparel PCE tracked at a $364 billion annual run rate in the first quarter of 2017.

A five percentage improvement in conversion, sharply below levels suggested by operators on enhanced in-store marketing technology, would suggest an $18 billion annual revenue opportunity.

Optimizing inventory Overstocks and out-of-stocks cost retailers $1.1 trillion globally in lost revenue, according to IHL Group.

In North America, the loss from overstocks in the region is estimated to cost retailers $123.4 billion annually and out-of-stocks $129.5 billion.

Managing these assets carefully is critical to a sector under secular pressure.

Goldman Sachs Global Investment Research 20 August 2, 2017 Profiles in Innovation

Who else is talking about the store of the future? We may be talking about the store of the future, but not many retailers are. And most that are have started only recently.

Walmart is starting We were struck when Walmart CEO Doug McMillon thrust mentions of some of the key topics in this report into his comments at the company’s annual shareholder meeting, a mammoth affair in a collegiate basketball arena where the firm shares extensive observations about its performance and market but rarely breaks new ground.

“Tech is changing what customers expect and how they shop. The next verse in our song is about them, our customers. They are the reason we have jobs. If we don’t get ahead of where they’re going they’ll choose to shop somewhere else. So where are they headed? They’ve always wanted the best prices, they want value for their hard earned money and they should and they want great merchandise, high quality, something new, something just flat-out awesome. And today more than ever customers expect, they deserve, to save time. Time is a currency like money.

They want an easy, enjoyable experience without friction that’s got to be hast. The historic trade-off between price and service doesn’t exist anymore. We’re creating a better shopping experience and a better work experience for ourselves as we strive to serve them.

For example, Tony started as an overnight stocker, now she manages the grocery department. We were in her store and she showed me how she drives modular accuracy and in stock with a new app we developed…

Personalization engineer using machine learning to monitor the weather and make sure we’re sending swim suits to FL and rain coats to Seattle at the right times.

Carlos Kirk is [an] IoT engineer. He’s experimenting with in store sensors to provide real time inventory and price changes.

Bowen Gong, data scientist for Sam’s, leads computer vision efforts that are aimed at eliminating the check-out process all together. We are experimenting and we are inventing.

We have tests going on with digital endless aisle shopping, automated pick up towers in stores, automated pick up stations in the parking lot, robotics and image analytics scanning aisles for outs, block chain for food safety and machine learning in our pricing systems to assist merchants.”

Who else? We recently searched for key terms used by large retailers to assess their focus on these topics. We reviewed the last 10-K and annual report filings, the last four earnings call transcripts, and the last analyst meeting transcript (2016-17, where applicable) for the ten largest retailers by market-cap (Walmart, Costco, Home Depot, Lowe’s, TJX, Target, , O’Reilly, ULTA, BBY) and two of the largest department store chains (Macy’s and Nordstrom) for the following phrases: “IOT,” “internet of things,” “internet-of-things,” “sensor,” “artificial intelligence,” “AI,” “virtual reality,” and “augmented reality.”

We did not find much.

The most substantive content came from Macy’s and Kroger.

Goldman Sachs Global Investment Research 21 August 2, 2017 Profiles in Innovation

Macy’s – pushing department stores forward Jeff Genette, newly named CEO of Macy’s said on the firm’s year-end 2016 conference call that “…we acknowledge that more experimentation with in-store technology and creative solutions for customers who are shopping in our physical stores is an opportunity for focused capital investments in 2017.”

Also very recently, at its June 2017 analyst meeting, when it was asked about its AI efforts.

Justin MacFarlane, Chief Strategy & innovation Officer:

“…there is a lot of noise out there in terms of artificial intelligence, machine learning. And the question we always get is how do you apply all of this to Macy's. And from my analytics team which is amazing to even say that there is [sic] analytics teams in retail these days. Five years ago, if you were to say there was a team of PhD data scientists at Macy's that are thinking about our most strategic problems, people would have thought you are crazy. And we've moved that very, very quickly. And the real power of data and data analytics is focusing those resources and that talent on the core business. So, when I talk about folks about our focus, it's pricing inventory and customer. And if our folks aren't working on one of those three areas, then we're working on the wrong thing. And from a pricing standpoint, obviously, a lot of conversation here about margin. We have an unbelievable amount of opportunity in terms of optimizing our pricing. And that's secondary to all of the things we talked about in terms of coupons and clearer value, but we are [indiscernible] with markdown optimization, but with a really strong data science team, we can now start to take existing systems and processes and infuse them with just amazing algorithms that will allow us to get a point here, a point there, make the right decision on a markdown at scale. So tremendous opportunity there.

From a customer standpoint, everything that Rich [Richard Lennox, Chief Marketing Officer] talked about today needs to be enabled by data. How do we target customers, how do we micro segment customers, how do we personalize a message to our customers? That all starts with analytics and good data. And we have come great strides over the last 18 months, and being able to really get to a point where we know our customer better than we ever have, and we have the capabilities to talk to her in ways that are much different than we've ever been able to do.

And then the last one is inventory. As my partner R.B. [Robert B. Harrison, Chief Omnichannel & Development Officer] talked about, and Patti [Patti Ongman, Chief Merchandise Planning Officer], how important is getting product in the right place at the right time, and even more importantly, as we start talking about BOPS and store fulfillment, making sure they we're optimizing that through data to make sure we're pulling product from the right place at the least cost, closest to the customer with the best speed. And we've infused a number of analytics and new algorithms in order to optimize all those processes. So, we like to be very practical, when we talk about machine learning and analytics and really focused on those three core strategies of customer, pricing and inventory.”

Robert B. Harrison:

“There's a number of places that where we've been using it. If you did follow our app, we had a voice recognition feature up on the app, about six months ago. We had it running for about six, nine months. And we actually at that point were getting a lower conversion than we got from this traditional interfaces. That said, I agree with the premise that I believe voice is where – voice recognition is going to be a key aspect. Chatbots are a fact of life in a lot of our interactions already. We have Messenger bots running on Facebook. We are in pilot with a large

Goldman Sachs Global Investment Research 22 August 2, 2017 Profiles in Innovation

technological vendor to take our chatbot on the customer service centers to call centers to another level. And as Justin mentioned, there are a number of places we're applying machine learning; the PFL, the preferred fulfillment logic that Justin mentioned is one, because we literally run every transaction we do through the PFL, at nanosecond time to go figure out where to get it and what we can learn from that, how do we optimize that customer experience. There are places in the pros right now. The product recommendation features on the site, where we believe you're applying machine learning to elevate and improve how we recognize you and customize the home page. So we're bought into it. Right now, it is one of the classic buzzwords and the trick is trying to figure out who you want to partner with and who you want to bet on is a right way to go. And to do that in a way that makes sense from a business and customer perspective. So, we will be there.”

Kroger: speaking up for Kroger, in its annual report, focused on it as well:

“Digitization of the Store: Another way we continue to push the boundaries with data and technology is through a series of initiatives at scale that, taken together, comprise what we believe may be one of the largest Internet of Things deployments in the world.

This includes our digital temperature monitoring deployment, which monitors and regulates temperatures in every refrigerated and frozen case in our stores. In addition to saving us money, freeing up our associates to take care of customers (rather than manually logging temperatures), and improving food safety by removing the potential for human error, this initiative is also giving us something of unrivaled value: even more data. Our operators and engineers can apply this new data toward process improvements, further cost savings initiatives, and perhaps even commercializing and selling this new technology to other retailers.

This is but one example of what Kroger’s in-house team of inventors and innovators can do by leveraging one of the most impressive digital labs available in the world: 2,800 actual store locations and a real retail environment to test and learn in.” Many other retailers are unwilling, or unable, to engage.

Goldman Sachs Global Investment Research 23 August 2, 2017 Profiles in Innovation

Interview with…futurist Steve Brown

Steve Brown, also known as the Bald Futurist, is a futurist, speaker, consultant, and executive coach, specializing on the future of computing, with over 30 years of experience. He founded Possibility and Purpose, LLC (BaldFuturist.com). Brown is the former futurist and chief evangelist at Intel Corporation.

Matthew Fassler: How has the retail industry typically thought going to need to double down on that experience and use technology about investing in technology? where it makes sense to deliver a better experience to consumers, and their higher levels of personalization, customization of product Steve Brown: Well, retail is notoriously cost-focused. They’re very and service, new ways to discover products to delight and surprise good at managing their cost, that’s how they’re successful, so they them, wrapping service around products. tend to see everything as a cost center, and they have tended to look at IT investments as what do I need to do to keep the business And on the other side, the non-experiential retailers, who are selling running, rather than an opportunity to disrupt, to bring new their non-differentiated, more commodity goods, are going to have to innovation, to create new value. Typically, IT is pegged as 1%-1.5% invest in new models. So, for fulfillment centers, start looking at of sales for a retailer, and they need to shift that mindset, from IT as ambient shopping, automated delivery, new fulfillment models. I think plumbing, to IT as a competitive weapon. we’re going to see a big disruption as the two parts of retail split apart. And we may even see a disaggregation of browse and fulfill in MF: If we think about the changes that we’ve seen in the the shopper journey. retailing landscape, how have retailers’ imperatives to invest changed with the emersions of ecommerce and with the threat ------to the retail store? “There are a lot of innovations out there that I SB: Well, our threat is growing. I mean Amazon is growing, investing heavily, not just online now but also started to invest in physical would call shiny objects. They look sexy, and stores and leveraging their technology infrastructure and their they seem like they might be a cool idea and a technology know-how to reinvent the physical store. Walmart’s great consumer experience, but ultimately, investing heavily, too. I think the other big retail brands are looking at that and they’re just standing on sidelines to some extent. It’s a bit they don’t deliver business results.” like Nokia when they were watching Apple and Google eat their ------lunch in the phone business by playing a totally different game. So, I think to some extent retailers are not really sure what to do. MF: If you can harness your IT expertise, your experience and training as an engineer and working with big tech, what are the At the same time, consumers are getting more and more more interesting technology investment opportunities that comfortable with online purchases—in certain categories. In other retailers are confronting today? What is emerging that warrants categories, they’re not comfortable with that yet. And I think online, significant investment in your view? And conversely, where do particularly mobile, is where a vast majority of shopper journeys you see retailers investing against dead-end outcomes? start. Most shopper journeys end in brick and mortar retail, but that transition is the danger zone. And capturing those people and SB: Anything that accelerates the fulfillment process, using making sure they come to brick and mortar retail is the biggest , is going to help, whether you’re an experiential retailer or challenge that these retailers have. a commodity retailer. Voice platforms, chat box, they are definitely a way that consumers, particularly millennials, and Gen Z after that My fear for them is that this pressure is not just going to continue to want to interact with brands, so every retailer needs a voice strategy. build from Amazon and Walmart, as those two heavyweights continue to make the investments that they’re able to make because Predictive and prescriptive analytics to help you manage your of their scale, but that it gets worse. It gets worse because of the business better and to understand customers better. In-store emergence of low-cost rapid delivery. That removes the immediacy sensing, to feed those analytics, so RFID, Wifi, cameras, indoor barrier for ecommerce. People go to stores, sometimes, because location type services. I think smart stores and smart checkouts to they want a product right now and they’re willing to wait two days for reduce the friction in the buying process are going to be key. it to be delivered to their homes. All of this needs to be secured so end-to-end data security is Once we move to an era of low cost, rapid delivery, I’m talking within something that they should all be looking to deploy now . one or two hours, that really cranks up the pressure for traditional On the emerging side, I think ambient shopping—it’s called ambient retailers. So, what I see happening here is what I call the great shopping, the fourth channel, smart home connection, but using the bifurcation of retail, where we’re going to see retail split down the home as the next storefront. So, the ability to sense what is needed middle, and split into two buckets. One of experiential retail, and one in the home. And that can be anything from, sort of , kind of being commodity consumption. An experience brand, they’re

Goldman Sachs Global Investment Research 24 August 2, 2017 Profiles in Innovation

Amazon Dash type products, Haiku [Homes], to energy sensors. Those technologies are going to become a really interesting have observed, such as sensors that yield observations that are feedback loop. To remove the cognitive load for consumers, who ultimately interpreted via AI and via machine learning, seem to don’t want to have to think about certain purchases. come with hefty price tags. And it seems quite costly with today’s pricing structures to populate a single store, let alone a AR and VR shopping in the home, that’s something that’s emerging, chain of stores, with the kind of coverage that would yield the it’s going to be with us fairly rapidly, so in the next couple of years optimal results. How do companies think about the economics that will hit the mainstream. Shifting to a model with the store as a of some of these innovations today? And how do you see the showroom, so making it easy to go into showroom and then make cost curves evolving? Where might we get scale and where purchases online from that showroom. might they remain prohibitive?

Wearables, for customer service agents, in-store robots, which is SB: I think you’re right. There are a lot of innovations out there that I going to be appropriate in certain categories, I think. would call shiny objects. They look sexy, and they seem like they And then delivery automation. If you look at what Starship might be a cool idea and a great consumer experience, but Technology is doing, they are looking to ultimately deliver product to ultimately, they don’t deliver business results. It’s all an ROI analysis. your front door for a dollar. That is incredibly disruptive, and anybody A new technology must do one of two things: increase revenue by that’s not looking at that needs to figure out how to either develop increasing traffic, conversion, basket, or affinity, or reduce costs – that capability or partner with someone that is developing that inventory costs, labor costs, real estate costs. If it doesn’t move any capability. Those would be the things I would look at. of those dials, then it doesn’t have a place in the store.

MF: We’ve noted a number of companies working on robotics, It could also be worthwhile if it accelerates, disintermediates, or but other than this success that Amazon seems to have had removes an existing business process. If a new technology doesn’t with Kiva, driving its distribution, the applications in-store, and help do one of those things you shouldn’t be looking to deploy it. The even on the back-end for many retailers seem limited so far. way you find that out is a three-step approach. How do robotics play out in your view, in real life, considering The first is to build a full test store in a lab environment. So, it’s not where the technology stands today, and what role would you deployed for the public, but it allows you to evaluate the technology say that robotics end up having in the brick and mortar space in feasibility and eliminate any usability issues, and you test it with real particular? people but in a controlled environment.

SB: I think the role of robotics depends a bit on the store Once you’ve got that rung out, and only once you’ve got that rung environment. If you are a retailer that’s selling apparel, it’s probably out, then you deploy it in test stores, One-to-ten stores, and the key not something you want to deploy on a very cramped floor, and it there is to evaluate the business impact in a real-world deployment, doesn’t really help solve many of your business problems. If you’re a and find any issues that relate it to being an uncontrolled greeter or a guide to help customers find what they’re looking for environment, but really it is to look at the business analysis on those makes a lot of sense. Similarly, in grocery I think robots are stock stores and see if you get an uplift in traffic conversion, basket affinity, takers to do all that stock taking, which can be about 40% of the time and/or if it helps you control your costs. And then you do an ROI spent by the workers on the shop floor, that can make sense, so if calculation, and it’s as simple as that. you look at what Bossa Nova is doing in that area, they might be interesting. Just to do the manual labor, to help people carry MF: What are we missing as an ecosystem as relates to the role purchases to their car, for example, might be interesting. That’s the of mobile? This has been the essential element of the role that they’re going to play in front of house. omnichannel experience. How is it evolving?

Back of house, in the stock room and fulfillment centers, of course it SB: We’re at the height of the mobile era right now. Mobile is the makes sense. Amazon is very smart to buy Kiva and take them off remote control of modern life and it’s not going away anytime soon, . but I think how we use it is going to change a lot. We have lived in the era of the “GUI,” the graphical user interface, even though it’s I think the other area that is interesting for robots is delivery. been updated to be touch now, for thirty years. We’re about to move Whether it’s delivery robots, that go by ground, like Starship to the era of the “VUI,” the voice user interface. As we see AI in the technologies is looking at, or delivery drones. And delivery drones background improve the capability of virtual personal assistance – don’t have to do the last few miles. They may be paired with an and I’m talking about Siri and Facebook, Cortana, Alexa, Google autonomous delivery truck to literally do the last 50 feet from the Assistant – as those platforms improve— [Alexa speaking in delivery truck to drop things onto the porch. So those seem to be the background] sorry my Alexa in the background just picked up me roles that robots may play, really augmentation in the store is going talking to them – retailers are going to find that these voice platforms to come more in digital form than physical artificial intelligence form, are a key part of their strategy to interact with customers. So, which is what a robot is. smartphones are going to be used not for touching as much, but MF: How do companies weigh the cost of innovation versus the instead, for talking to. There’ll be a portal for a voice conversation benefits of innovation? Some of the new technologies that we with brands.

Goldman Sachs Global Investment Research 25 August 2, 2017 Profiles in Innovation

A note on the importance of scale Scale has been increasingly critical to retail for decades, primarily because technology enabled increasingly larger enterprises that brought powerful economies to purchasing and marketing, and more recently because it is highly challenging to run a viable ecommerce business without serving a national stage base from the outset, given fixed costs associated with establishing awareness, running an innovating first-class online store-front, and running an efficient DC. The imperative to invest in in-store sensor infrastructure, state-of-the-art AR/VR capabilities where relevant, and strong data analytics will add another layer of cost and competitiveness to the sector, advantaging larger firms – provided they are forward thinking. It is no accident that some of the most evident focus from retailers has come from Walmart, the nation’s largest retailer, and Macy’s, the largest department store firm.

Goldman Sachs Global Investment Research 26 • Beacons: Once feeding sales information to customers, now feeding data to retailers (p. 28)

• Computer Vision: Putting fresh eyes on customer and product flows (p. 31)

• RFID: Improving inventory management, from allocation to automated replenishment (p. 32)

• Artificial Intelligence: Making sense of an abundance of data to optimize resource allocation and store-level execution (p. 36)

August 2, 2017 Profiles in Innovation

BEACONSThe role of IOT How does a store connect with customers on the go? Gathering intel with the help of Bluetooth

A brief introduction to beacons In an effort to gain more connectivity with consumers in-store, some retailers have installed beacon transmitters within their footprints. The beacons themselves are small objects often placed at the entrance of or within stores that sends signals to communicate with phones within a given radius. The typical original use for beacons was to connect with nearby customers who had the company’s app downloaded onto their phones by pushing notifications about promotions to entice them to walk into the store, or to ping customers with promotions once they had entered the store, with precise offers often linked to their precise location in the box at that moment. Expanded-use cases also include providing the customer with real-time data, such as reviews or product specs, in addition to helping customers locate a specific good. Beacons, though, are evolving into sensors, as these same devices, along with others already embedded in the store ecosystem – lighting, security cameras – return critical information to retailers, feeding their analytical engines, without the need for witting, deliberate participation by consumers. Below we detail the technology behind beacons, challenges that the rollout has faced, and the latest beacon endeavors.

What is beacon technology? And how is it morphing to a broader sensor strategy? Beacons gained popularity with retailers by leveraging low-energy Bluetooth (“BLE”) technology, in most instances, and similar systems have grown to leverage Wi-Fi and MAC signal compatibility as well. Utilizing BLE, which has modest energy requirements, the signal from beacons sends information and avoids draining phone batteries as GPS- enabled apps are prone to do. The Bluetooth signal is received by the connected consumer’s device and the relevant data is captured within the app on the receiving end and not sent back to the beacon. Growth in the market was spurred by the launch of Apple’s iBeacon in 2013, followed by Google’s entrance into the market in 2015 with its Eddystone product. Beyond these two major players there is an entire ecosystem of beacon hardware and software companies, highlighted on page 10.

Bluetooth is ideal for short-distance, wireless communication between two devices, while Wi-Fi is a much broader system carrying full streaming internet from a hub to the device. We also note that, in its basic form, Bluetooth does not provide any demographic customer information, but that each device has a specific signature that identifies a phone, and hence its user (assuming only one per device) on an anonymous basis. There are privacy laws around individualized data storage on customers; anonymity is important given privacy restrictions. As soon as a retailer begins to store personalized data or track customer behavior without their express consent via an “opt-in,” there are significant legal and compliance hurdles to overcome around protecting it, including who has internal access to non-anonymous data and the level of detail available.

One of the challenges with using BLE is that consumers must have Bluetooth enabled on their devices, which many have turned off. As a result many beacons now leverage other means of communication, such as Wi-Fi, that can ping to consumer phones differently. Unlike BLE beacons, Wi-Fi systems allow data to travel in two directions. Wi-Fi systems can both push notifications out to consumers and receive data back. Additionally the use of Wi-Fi systems removes the requirement for the customer to have a retailer’s app

Goldman Sachs Global Investment Research 28 August 2, 2017 Profiles in Innovation

downloaded to gather data on the user, though it is still necessary to enable push notifications. Due to wider communication range, Wi-Fi systems also require less physical beacon density than BLE systems. One barrier to entry for some Wi-Fi systems is that customers may need to actively log on to the signal, vs. passive / automatic connections depending on the system settings. “Guest-wifi” is being leveraged as an additional way to gather data on a consumer, if they are willing to actively log in, that provides a rich cache of data on the customer’s physical path during their time in the store, but also provides broader data from their mobile device for analysis. This could include which websites are accessed while the customer is in store or even provide broader access to the content of the user’s phone.

Proximity sensors are increasingly ubiquitous. Unacast, a firm that claims to operate the largest network of beacon and proximity data, expects 500mn proximity sensors globally by 2021, up 80-fold from approximately 6mn in 1Q17.

We outline the broader spectrum of communication technologies below, but note that most of the offerings we have seen are based in BLE or Wi-Fi.

Exhibit 11: There are multiple technology platforms for these networks and data gathering

Accuracy Presence Position Power Range Technology Identification (distance) detection type use (meters) Disadvantages Ultra Wideband Yes cm-dm Yes Absolute Low 1 - 50 Signal can be blocked by metal objects Wi-Fi Yes m Yes Absolute High 1 - 50 Use if ISM band - interference Bluetooth Yes m Yes Absolute Low 1 - 20 Use if ISM band - interference, low range RFID Yes dm-m Yes Absolute Low 1 - 50 Low range and small coverage, unsecure communication Camera No mm-dm Yes Absolute High 1 - 10 Requires big computing power, requires line of sight Laser No mm-dm Yes Relative High 1 - 5 Requires line of sight, provides only relative positioning Infrared No m Yes Absolute Low 1 - 5 Requires line of site, can easily be blocked by opaque objects Ultra Sound No cm-dm Yes Absolute Low 1 - 10 Susceptible to acoustic noise

Source: Sewio.com

Many of these sensing technologies enable geofencing for retailers as well, defining a specific physical area where beacons are active, such as the boundaries of the store or the areas around the exterior. This allows retailers to target certain physical areas with certain promotions or recognize when a potential customer enters the parking lot by picking up the signal in advance of entering the store.

The evolution of beacons use cases Beacons were originally installed to spur interaction with nearby customers, deploying targeted marketing efforts with promotions or product info. While these efforts had some success in driving top-line sales, the model faced multiple challenges.

One of the biggest hurdles was that in order for a beacon to ping a customer, that customer would have to already have the app downloaded on their phone that the beacon could connect to. Google launched an app to help facilitate the contact between retailers and customers for users on its Android platform. Rather than needing to have an individual retailer’s app open, by using the “Nearby” functionality consumers are able to use a single interface for beacons across all retailers in the vicinity. Still, this only addressed the app requirement for Android users, leaving Apple and other device owners reliant on prior app downloads.

In order for the beacon to communicate with a mobile device, the phone must have its Bluetooth enabled, which many consumers may leave off.

Goldman Sachs Global Investment Research 29 August 2, 2017 Profiles in Innovation

Aside from the technical requirements, there was also backlash from customers who found the inbound messages to be annoying or did not like having their phones pinged. A Localytics customer survey found that over 50% of consumers found push notifications to be an “annoying distraction”, and one company stated they found that if a customer gets a push notification upon entering a store that there’s a 300% chance they delete the app if they receive another notification from the retailer again in the next 30 minutes.

Furthermore, consumers have not yet developed the behavior to check their phones during a shopping experience for shopping related-content. While they may check their phone for messages or place a call, the drive to shop with ones phone for promotions did not catch on with consumers.

Recognizing the many challenges to the push-notification marketing approach with beacons, retailers have shifted to a different use case with beacons in stores, specifically data collection through mobile fingerprinting. Multiple companies now offer services that will track customer phone signals, often through Wi-Fi, as they shop a store, capturing data that can be utilized by the retailer in business analytics rather than direct customer exchanges.

Goldman Sachs Global Investment Research 30 August 2, 2017 Profiles in Innovation

COMPUTERA brief introduction VISION to beacons Fresh eyes on flows and product

The role of computer vision The British Machine Vision Association and Society for Pattern Recognition defines computer vision as “concerned with the automatic extraction, analysis and understanding of useful information from a single image or a sequence of images. It involves the development of a theoretical and algorithmic basis to achieve automatic visual understanding.” In other words, traditional images yield enhanced results through advanced analytics.

These are now combined with photography, video, and, in some instances, RFID technology, to present a comprehensive overview of the flow of people – consumers and employees – and product through a store.

Firms can track store shelves to assess in-stocks, planogram compliance, and sales velocity; observe consumer traffic; and, deploy image capture for the purpose of emotion recognition (see more on page 40).

Goldman Sachs Global Investment Research 31 August 2, 2017 Profiles in Innovation

AMZN’s Amazon Go store, still in beta and detailed in our vignette on pages ____-__, captures images on a high-frequency basis, using a database of products for tracking RFID inventory, and matching a consumer with the image captured upon entry/exit to substitute Keeping thefor the shelves checkout processstocked

The role of RFID Retailers have turned to radio frequency identification (RFID) to enhance their ability to localize, source, and track inventory at the SKU or item level.

RFID technology allows for the automated identification of objects through a digital chip that contains item-level information, which in turn can include an item’s serial number, price, and product information (i.e., size, color, etc.).

The RFID chip is in many ways similar to a barcode, but it is differentiated in its ability to transmit its stored data, wirelessly, to an antenna, which can then be sent to a larger server, enabling retailers to track inventory, and its movement, real time. Unlike a barcode, RFID does not require “line of sight” technology, so its signal can be read as long as it is within the range of a reader, as opposed to a barcode that must be oriented to a scanner. RFID diminishes the need for inventory tracking and the labor associated with taking physical inventory.

RFID is most commonly utilized by apparel retailers given both the cost of a tag and the unpredictable nature of apparel demand; constantly evolving fashion trends and changing weather trends require high product demand accuracy when planning inventory levels.

What does RFID do? RFID follows every item in its journey though the supply chain to point of purchase. An RFID tag is attached to each SKU before a product shipment leaves the manufacturer’s site. Each RFID tag typically contains an Electronic Product Code (EPC), or a unique number that identifies a specific item in the supply chain that can include a product’s serial number, its manufacturer, its SKU, and any relevant information on the product (e.g., size, color, etc.). The EPC allows retailers to track an item through each stage of the supply chain across different manufacturers and vendors.

When an item arrives at a distribution center, it passes by an RFID reader, which sends a signal to a larger network that can provide retailers with a live view of product availability at the DC. This real-time tracking can enhance visibility beyond the store’s backroom. When the product arrives at the store, the item’s record is updated automatically, and its location within the store can be monitored real-time as its RFID tag can send signals to antennae located throughout the store. This localization allows sales associates to adjust any misplaced items or quickly find an item for any reason (i.e., in-store or online customer inquiry). Additionally, RFID-friendly antennae can be implemented in fitting rooms, providing further insight on in-stock levels, while also measuring fitting room conversion rates (i.e. how many items that were tried on were ultimately purchased). RFID tags follow each product to checkout, automatically updating inventory records as each individual item is sold.

The most sophisticated RFID systems can automate replenishment using predictive analytics. Through machine learning, these systems assess a live feed of inventory flow, predicting future stock needs based on current stock levels and historical trends. These systems can also discern how to most optimally allocate the available stock in a DC based on different store inventory levels. The end goal of these technologies is to generate the most accurate order forecasts driven by granular data rather than human guesswork. Blue

Goldman Sachs Global Investment Research 32 August 2, 2017 Profiles in Innovation

Yonder, one of the startups using machine learning to automate replenishment, has reduced out-of-stock rates for its customers by up to 80%.

Origins RFID originated from a predecessor technology deployed in World War II to identify incoming aircraft. Both the Allies and the Germans used radio wave frequency during night missions to determine whether oncoming planes were allied or enemy. The British established a system titled Early Identification Friend or Foe (IFF) in which each British plane would have a transmitter that could receive coded identification signals over radio.

The 1950s and 1960s saw the continued evolution of RFID technology, with the first tags developed for use with anti-theft systems. These electronic tags contained a coiled metal antenna in the label that would send a current to a concealed receiver in a store’s doorway. The label remained “on” as long as the tagged item was still in the store and would be turned off when the item was purchased. If an individual tried to remove the item from the store before the tag was turned off, the receiver would pick up the signal from the label and trigger an alarm. This technology is still in use today, though higher-frequency RFID tags have been developed since then and the use cases extend far beyond just security.

In the 1990s, IBM began developing ultra-high frequency technology that was designed to extend a chip’s reading range and accelerate the flow of data. IBM reached out to Walmart to pilot the technology, but it never made it past the initial stages as it was too expensive to implement on a larger scale at that time. IBM eventually sold its ultra-high frequency RFID patents to Intermec, which became one of the leading providers of RFID technology.

A major breakthrough in RFID’s history occurred at the Auto-ID Center at MIT in 1999, when the Electronic Product Code (EPC) was created. The Auto-ID Center was a collaboration between MIT and over 100 global corporations to develop RFID tags embedded with individualized EPCs, which would limit the information carried to just a serial number and enhance visibility across different suppliers and manufacturers. These codes would be stored on a larger network with all of the relevant product information, as opposed to overloading each tag with all the information on that product. This simplification process lowered the cost of tags and is the primary technology used in chips by the majority of retailers today. The initiative saw significant involvement from over 100 global corporations as well as significant vendors in the RFID space, but it was still very early stages and was difficult to scale at this point.

In 2003, Walmart re-entered the RFID arena, with CIO Linda Dillman spearheading an effort to embed RFID tags on all goods from Walmart’s top 100 suppliers. Ms. Dillman was on the forefront of attempting to utilize RFID technology to drive efficiencies in Walmart’s supply chain at an unprecedented scale. The mandate generated resistance from suppliers, as Walmart required the suppliers to pay for both the tags and the application of the tags. For suppliers distributing lower-margin product, the costs were substantial: AMR Research estimated that the cost of tagging product for a supplier shipping 50 million cases per year would have been anywhere from $13 million to $23 million. Additionally, the new initiative meant that Walmart’s suppliers had to manage a separate flow of inventory tagged with RFID, in addition to its main inventory flow for the rest of its retailers. Ultimately, the project faded.

Current state of the industry More recently, as the shift to omnichannel creates additional supply chain complexities and requires higher levels of stock visibility, RFID has become a more viable investment for retailers. Ramps in chip production have lowered the cost of implementation, and established retailers such as Macy’s and Inditex have reported substantial efficiencies realized after installing the technology. A survey from the Platt Retail Institute found that, in the Macy’s Women’s Shoe Department, RFID improved inventory accuracy and order

Goldman Sachs Global Investment Research 33 August 2, 2017 Profiles in Innovation

fulfillment tracked higher for RFID-enabled merchandise. Given the success it has had so far, Macy’s is working toward RFID implementation in suitable categories to be completed in the relatively near future.

The RFID industry remains relatively fragmented as a whole, though there are some players that have been more relevant to the evolution of RFID in retail. The majority of these vendors not only supply the hardware, such as chips, inlays, and readers, but they also offer services and software platforms that can tie together product data to help retailers manage a connected platform. In terms of chip producers, NXP Semiconductor (NXPI, covered by Toshiya Hari) is a leader in RFID tags and labels, with strength in NFC (near field communication) readers, a subset of RFID which can also perform peer-to-peer communications. Alien Technology and Impinj are most well-known for supplying UHF Gen 2 RFID chips, the most popular chips used in retail. Avery Dennison is an industry leader in supplying inlays, or smart labels, which take RFID chips that store product information and attach them to an anetenna that transmits the radio frequency signal. Tyco Retail Systems has also had a notable presence in the RFID space, partnering with Inditex and Macy’s to implement chain-wide RFID systems using RFID-enabled hard tags that double as security tags. Zebra Technologies is also an enabler of RFID in retail, with RFID- reading scanners and sleds as well as printers that activate the technology on labels (note they don’t manufacture RFID tags themselves).

Use case: Tracking inventory One of the biggest challenges that retailers currently face – and the most popular use case for RFID – is inventory management. According to RFID inlay manufacturer Avery Dennison, average inventory accuracy currently tracks at about 65% for most retailers, but with RFID- embedded tags, that number increases to 95%. In order to combat the unpredictability of inventory management and to ensure higher levels of accuracy, retailers are increasingly turning to RFID-based inventory management systems.

Although the majority of retailers have yet to fully implement RFID on a large scale, large retailers such as Inditex and Macy’s have specifically cited the benefits they have seen from RFID. Last year, Inditex CEO Pablo Isla noted that RFID implementation at Zara has been the most significant change to store management the company has seen in decades as it has eased the flow of store operations through faster and more efficient replenishment. Inditex has invested over one billion euros over the past four years in RFID technology and is working to deploy it in all Zara stores. H&M has also announced plans to implement RFID technology as well.

Obstacles Challenges to implementing RFID include the enormity of changing inventory monitoring practices, and running redundant systems. A notable example of a transition gone wrong was the failed RFID launch at JC Penney in 2013. The company ultimately abandoned its effort to implement an RFID-based inventory management system given cost pressures, but in preparing for the switch to RFID, it had removed security tags from merchandise due to signal interference between the two systems. When JC Penney gave up the RFID project, half of its merchandise lacked security tags, making the retailer a prime target for theft.

Cost thus remains a critical constraint in the implementation of RFID, and has been the most commonly cited deterrent across the retail landscape. Not only do the chips themselves require an investment, but also implementing the overall infrastructure requires a substantial commitment in terms of capital deployment and time invested, with many retailers remaining unsure that the benefits will outweigh the costs.

Notably, however, cost continues to come down over time as retailers innovate and chip manufacturers realize efficiencies in the production process. A major breakthrough in the progression of RFID adoption in retail occurred when Zara chose to put the RFID chip in the

Goldman Sachs Global Investment Research 34 August 2, 2017 Profiles in Innovation

larger security tags already attached to each individual item instead of adding a new tag for the RFID chip. This allowed Zara to re-use its RFID chips rather than throwing them away after each use: security tags were removed at checkout and sent back to the chip manufacturer where they were recertified, tested, and sent back to the apparel manufacturer to re-enter the supply chain. Additionally, the cost of RFID chips continues to fall as more chips are produced and an increasing number of companies adopt the technology. EPC Global, an initiative created to drive the widespread adoption of RFID, has stated that chip costs have dropped notably over the past few years and will continue to fall as production ramps. The initiative’s goal is to drive down the cost to 5 cents per chip, down from its current estimated cost of 7 to 15 cents. In its pitch to retailers, RFID chip manufacturer Avery Dennison highlights that the investment usually covers itself within six months of implementation. Current outlook

The rate of adoption across retailers is accelerating: the RFID Lab at Auburn University reported that the rate of RFID adoption among the top 100 US apparel retailers tracked at 32% in 2016 vs. 23% the year prior. Even within apparel, there is room for further penetration as only 15% of the total addressable market for apparel used RFID in 2016, according to IDTechEx Research. The RFID industry also has room to grow outside apparel. Kurt Salmon’s 2016 RFID in Retail study found that categories such as consumer electronics and sporting goods are turning to RFID to more optimally manage inventory and fulfillment in order to combat the competition from online players. As RFID costs dwindle and brick & mortar retailers realize the importance of store optimization and dynamic omnichannel fulfillment, RFID adoption should continue to gain traction.

Goldman Sachs Global Investment Research 35 August 2, 2017 Profiles in Innovation

ARTIFICIAL INTELLIGENCE

Optimizing store execution

Making sense of it all; AI drive processing power Large retailers operate broadly distributed businesses, with massive customer bases, endless permutations of customers, SKUs, prices, locales, weather conditions, and competitive inputs to consider when evaluating drivers of sales, share, and financial returns. This data can be tough to capture, and it is near-impossible to analyze – without AI, which enables speedier processing of outsized amounts of data, with the analytical process helping to suggest the questions that are asked, as well as to decode the outcomes.

For more on AI’s There are universal use cases in retail that transcend sector, and that within retail transcend many use cases, see distribution channel. Broadly speaking, we see potential to reduce the commitment of labor our Profiles in for simple processing roles – companies in our coverage have spoken to substantial Innovation report opportunities to reduce their headcount in basic financial functions like audit and accounts published November receivable. In the November 2016 Profiles in Innovation report, “Artificial Intelligence,” we 14, 2016. focused on specific use cases for retail including inventory management, store locations, recommendation engines, customer support, demand prediction, and price optimization. Use cases included reducing costs for merchants and optimizing prices. Most of these practices, with the exception of store location, relate to efforts that are equally applicable across direct and store-based channels.

But there is another layer of use cases – optimizing execution at the store level. Any given public chain embodies millions of permutations of SKUs (in the hundreds, thousands, or ten-of-thousands), stores (from a few dozen to several thousand), employees (from a handful of boxes to hundreds), across different times of day, at different times of year, interacting with external stimuli – weather, evolving local competition, changes in local management. AI firm Blue Yonder recently observed that data for a typical supermarket chain is 20bn rows long, presenting a daunting amalgamation of information that can be interpreted only through enhanced processing capacity.

The industry is now able simultaneously to capture these data points – with the IOT and RFID technology described above, as well as legacy means such has photography and video, and, by deploying AI resources, derive prescribed outcomes to help it manage the business.

Exhibit 12: AI empowers analytics

IOT Sensors AI Rapid Timely data Photography Rapid interpretation of capture Video data & ability to offer Feedback and upload Wireless prescriptive solutions Loop

Source: Goldman Sachs Global Investment Research

Goldman Sachs Global Investment Research 36 August 2, 2017 Profiles in Innovation

Defining terms AI: The leap from computing built on the foundation of humans telling computers how to act, to computing built on the foundation of computers learning how to act. Also, it is the science of simulating intelligent behavior in computers. It entails enabling computers to exhibit human-like behavioral traits including knowledge, reasoning, common sense, learning, and decision making.

Machine learning: Algorithms that learn from examples and experience (i.e., data sets) rather than relying on hard-coded and predefined rules. In other words, rather than a developer telling a program how to distinguish between an apple and an orange, an algorithm is fed data (“trained”) and learns on its own how to distinguish between an apple and an orange.

Deep learning: A sub-set of machine learning. In most traditional machine learning approaches, features (i.e., the inputs or attributes that may be predictive) are designed by humans. Feature engineering is a bottleneck and requires significant expertise. In unsupervised deep learning, the important features are not predefined by humans, but learned and created by the algorithm.

Furthermore, there has been massive growth in the amount of unstructured data being created by the increasingly ubiquitous connected devices, machines, and systems globally. Neural networks become more effective the more data that they have, meaning that as the amount of data increases the number of problems that machine learning can solve using that data increases. Mobile, IoT, and maturation of low cost data storage and processing technologies (often in the cloud) has created massive growth in the number, size, and structure of available data sets.

Goldman Sachs Global Investment Research 37 August 2, 2017 Profiles in Innovation

Blue Yonder

The opportunity Retailers possess increasingly massive volumes of information, some of which has long sat at their fingertips, waiting to be deployed, and some of which has recently started being gathered by the sensors described in this report. Moreover, retailers have the urgency to optimize operating performance in the face of intensifying competitive pressures, both from online entrants and from disruptive brick & mortar formats (i.e. hard discounters Aldi and Lidl accelerating US expansion).

The solution Blue Yonder deploys AI software to process countless decisions across multiple arrays of variables, focusing on inventory optimization and on pricing. The firm forecasts demand and necessary supply by assessing historical data (from the retailer), competitive locations and pricing (captured from the marketplace), and public domain data sets such as holiday calendars and weather forecasts. KPIs that typically benefit here include in-stocks/out-of-stocks, sell-through, and gross margin rate. The product is offered as a service, with the magnitude of the fee reflecting the magnitude of the solution, the SKU count, and, in some instances, the success of implementation, with Blue Yonder sharing in savings attributable to the rollout.

The company indicates that, stepping into Morrison’s, which had the opportunity for 13 million automated ordering decisions a day, the company generated a 30% reductions in shelf life gaps, a +2.9% yoy increase in sales, and a reduction in storage space in the back of the house. It indicates that for another (unnamed) lifestyle retailer, predictive shipping calculations on 2.2mn items, moved 5-7 day delivery to 1-2 days, increasing end customer satisfaction and driving sales.

The challenges Not all retailers have the data sets to contribute to the database to give it the optimal historical context. Moreover, capturing recommendations is not sufficient; retailers need to be willing and able to implement them on a massive scale.

Goldman Sachs Global Investment Research 38 Business Intelligence (p. 40) • Smart shelves & floors • Video analytics • Emotion recognition • Staffing • Smart pricing

Marketing (p. 45) • Smart dressing rooms • Digital signage

August 2, 2017 Profiles in Innovation

BI-focusedBUSINESS applications INTELLIGENCE Applications

Smart shelves Firms such as TRAX Image Recognition cater to the food and grocery business by using set cameras to take periodic photos – the frequency is based on the standards of the client, either a retailer or a brand – and transmit the image to a cloud-based backend system, where the images are compared to a product catalog and recognized. The software can track planogram compliance, providing remedies ranging from periodic ranking of departments, stores, groups of stores, or any combination of displays, or providing real- time feedback to the associate in the aisles to fix the problem. For the grocery space, the SKU count is high, average item value – and gross profit – is low, and consumer traffic count is high, such that monitoring item-by-item or consumer-by-consumer interactions is cumbersome. This effort aids inventory efficiency and reduces stock-outs, improving working capital utilization, aiding sales, and enhancing margin (given more accurate decisions on markdowns/clearance). Moreover, it reduces the amount of labor deployed by both retailers and brands to ensure in-stock compliance. Finally, it enables retailers to capture analysis on changes to sales resulting from product placement – placement of competing brands, or at different shelf heights – by tracking observed changes in shelf position or facings with changes in top-line results, ensuring that any measured outcomes in fact relate to differences in product display.

Diving even deeper, TRAX can move from the “smart shelf” to the “smart fridge.” Cooler- mounted cameras are activated by the opening and closing of the cooler door. The cameras stop recording as the door closes. Once the motion has been detected, the cameras begin to capture a series of images. The best images are uploaded to the cloud for analysis, with reports accessible online to designees at the retailer or brand.

Powershelf provides some of these features, adding temperature monitoring to the mix.

Smart floors Scanalytics is a company based around floor sensor technology (known as “solesensors”) and the accompanying dashboard that form a hardware/software package for a pay-as-you- go subscription. It can be used to monitor foot traffic, analyze consumer behavior and report back with actionable insights.

The company leverages paper-thin mats that can be used on hardwood floors, under carpet or outdoors to measure footstep volume and direction in the store. The mats measure two feet by two feet each and can be linked to increase measurement accuracy for large spaces, creating a network area that expands to cover the specific area desired.

The mats function essentially as a touchscreen on the floor, collecting anonymous impressions using pressure sensitive sensors, which are then processed by the company’s analytics suite leveraging its algorithms to identify traffic patterns and behavior. Retailers are able to better understand shopper flow through the stores, occupancy, and engagement zones and then leverage that data to improve its marketing and operational strategies.

Goldman Sachs Global Investment Research 40 August 2, 2017 Profiles in Innovation

Video analytics Videomining, which started doing contract work for the Departments of Defense and Security on the campus of Pennsylvania State University, integrates mobile fingerprinting but has particular expertise in video. The firm, focusing on the grocery and convenience channels, and selling largely to brands (versus retailers), deploys AI and machine learning techniques to classify consumers into demographic groups, by age, gender, and ethnicity, and then tracks and understands behavior. It deploys anonymous tags on the person when they enter store and track their visit through checkout. Time- stamping and getting the transaction basket data from the retailer helps capture a complete journey for each customer. This work can be applied holistically, across the store, or via a deep dive on a single category.

Emotion recognition Technology firms have added an additional layer of complexity to video analytics that can read and process human emotion. This emotion recognition software allows retailers to go beyond tracking just demographic data and can layer in notions of customer sentiment and engagement to a video-driven dataset. Using computer vision and machine learning techniques, the software recognizes facial expressions and categorizes emotional responses to product on the shelf in real-time. Emotion detection is more accurate in measuring a customer’s immediate feelings about product presentation than a survey – the technology tracks raw emotion in real-time, without the delay of a post-purchase survey and without the subjectivity inherent in responding to a survey.

Emotion recognition technology utilizes the Facial Coding System (FACS), a taxonomy of facial expressions first published for psychological research in the late 1970s. The system categorizes facial expressions based on different combinations of facial muscle positions, initially used to better understand the relationship between emotion and facial expression. More recently, computer programmers have integrated FACS into facial coding software that can classify human emotional responses picked up through an image or on video.

When processing the image of an individual’s face, the software’s computer vision algorithms locate the main landmarks of the face (i.e., eyebrows, nose, mouth, etc.), and machine learning algorithms subsequently track the movements of the face, analyzing facial muscle positions. The software is built to recognize a facial expression and then categorize a customer’s emotional response to a product based on that facial expression. Because the technology runs off a coding system, the classification process can be anonymized and does not store images or personal information on the customer. Retailers are thus able to access real-time metrics on customer engagement and sentiment, which allows for a more comprehensive, accurate data set that can help retailers understand how to optimally place product and lay out the store.

Affectiva is a marketing solutions firm that utilizes emotion recognition technology to process consumer responses to advertisements and retail products. The company has developed an LCD shelf display called shelfPoint that contains eye-level sensors to track movements in the face. When a customer approaches the shelf, Affectiva’s emotion recognition technology processes any facial movements and can categorize a customer’s response in real-time. shelfPoint can also detect demographic data such as age, gender, and ethnicity, providing a comprehensive data set that retailers can leverage. The digital screen can also update pricing and promotional activity in real-time, allowing retailers to dynamically price product.

Reflektion, in addition to customzing online search, creates a personal profile proscribing likely preferences based on the consumer’s browsing and purchasing activity in store.

Goldman Sachs Global Investment Research 41 August 2, 2017 Profiles in Innovation

Staffing Percolata engages computer vision, mobile fingerprinting, and audio analytics to capture shopper counts, dwell time, frequency of visits, and conversion rate, and deploys this data against staffing insights to optimize labor scheduling. By integrating video and audio data points with mobile fingerprinting, all of which is captured through sensors, Percolata provides retailers with dwell time data on a customer-by-customer basis, which gives a more holistic picture on traffic dynamics. Additionally, Percolata’s sensors track video and audio cues that provide insight on customer occupancy rather than in-and-out counts, which is what most competitors track and typically provides less accurate results. Though the data is anonymized, it still allows retailers to see how individual customers are moving about the store so that they can optimize staffing to best address customer needs.

Smart pricing Store-based retailers are constrained by printed pricing signage, such that pricing is less dynamic than it can be online, and price changes are costly to market. Firms such as Powershelf offer pricing signage in its IOT solution, focusing on power efficiency, a gating factor for digital signage in the past, with the value proposition aided by the ability to use the associated equipment as sensors for the use cases described above. These digital shelf signs allow retailers to remotely change prices, reducing the need for manual labor involved in changing a printed label every time a price is changed.

To accompany digital signage, firms have developed software to dynamically price product in the most optimal way for the retailer. In determining the best price, price optimization software integrates a plethora of factors that can impact customer demand while also considering the impact on retail margins. The calculation incorporates competitor pricing, external factors such as day of the week, upcoming holidays, and weather to determine demand elasticity and then layers in SKU-based inventory levels and operating costs to derive the most optimal price for that product.

The software’s predictive capability is derived from machine learning techniques that process historical sales and pricing data alongside the factors described above. The software generates predictions as to how customers will respond to future price changes based on how customers have responded to such changes in the past. Companies such as Altierre, Blue Yonder, Celect, and Revionics offer this software to help retailers boost sales and margins through more informed decisions on pricing, promotions, and discounts.

Retailers optimizing price can see a boost to revenue that flows straight down to the operating income line. Blue Yonder has reported that after implementing price optimization techniques, several of its clients have seen a boost to revenue and profits above 5%. In “Sizing the Price – The power of pricing,” a 2012 study of 100 price optimization projects across industries, Deloitte found that on average clients saw an increase in revenue of about 3.2% and that the majority of clients saw the technology pay for itself within 1-2 years. By incorporating dynamic pricing and price optimization into their strategies, store- based retailers can thus better compete with the online space while also maintaining healthier margins and sell-through rates.

A role for social media Many of these tracking tools are now being deployed by social media enterprises, which are capitalizing on users’ mobile engagement to track their location, commerce, and, where relevant, the connectivity between their own sites’ marketing functions and consumer spending outcomes.

Goldman Sachs Global Investment Research 42 August 2, 2017 Profiles in Innovation

Snap and Facebook are both utilizing their apps to track consumers’ movements, helping retailers to understand customer movements and advertisers to understand the efficacy of their spend and campaigns. Google has done this for years, helping advertisers gauge store conversion relative to ad spend. This data is typically aggregated and shared on an anonymized basis.

Snap’s “Snap to Store” tool, offered to retail advertisers that satisfy a minimum ad spend threshold, enables retailers to track whether user posts during and about a storefront establishment yields visits from friends who viewed that “snap.” Snap tracks consumers using GPS, assuming consumers have not opted out of being tracked. Snap recently acquired location-based analytics and ad measurement startup Placed to aid in this effort.

Over the last two years Facebook has also focused on matching consumer locations to stores, triangulating within an ecosystem of GPS, beacons, WiFi, radio signals, and cell towers. The endgame here, too, is helping advertisers assess the effectiveness of their ad spend.

Goldman Sachs Global Investment Research 43 August 2, 2017 Profiles in Innovation

RetailNext

The opportunity Many store-based retailers lack data, and many of the retailers equipped with data have not optimized its aggregation, analysis, and packaging for presentation to their organizations. Retailers struggle with questions involving membership and loyalty programs – whether and how to implement them; how to incentivize consumers to join; privacy; the logistics of data capture; and, analytics across massive data streams.

The solution RetailNext, founded in 2007, offers an array of solutions aimed at developing more robust analytics. It vertically integrates its analytics engine (think SAP), and presentation (think Tableau). The firm has engaged with retailers in over 80 countries.

At a very high level, the firm is engaging in two types of workstreams:

 “Traffic 2.0,” its newest traffic tracking system, aimed at measuring traffic, conversion, and drivers of each. It works to assess not only count but “capture rate” of passers-by entering the store; identifying customers as new vs. repeat, and if repeat, how frequent of a shopper; duration of visit; and interaction with sales associates, along with correlating that interaction with sales outcomes.

 “Retail Labs,” involving the use of sensors, as detailed in this report, to capture data and enable A/B testing along the lines of assortment, product placement, and pricing.

Beyond these specific solutions, RetailNext is leveraging its infrastructure and relationships to work on mobile, loss prevention, and omnichannel marketing.

The challenges The firm is tripped up by large retailers moving slowly, even in the face of compelling financial returns; by poor IT infrastructure; and, occasionally, by associate interference, as it sheds incremental light on individual productivity.

Goldman Sachs Global Investment Research 44 August 2, 2017 Profiles in Innovation

Marketing-focusedMARKETING applications Applications

A number of brands and retailers are deploying enhanced fixtures and signage to influence purchasing behavior, with the bonus (and stealth) benefit of feeding their analytical capabilities. Most of these efforts are focused in bigger-ticket, higher-margin businesses that command the significant investment technology encouraging a sell across a directed collection of SKUs.

Retailers are using sensors and RFID technology to create ecosystems that engage the consumer, with little effort from that consumer beyond steps they would typically take while shopping in the store.

Consumers entering stores to shop those stores, and are not necessarily looking to re- enact the steps associated with online purchase – i.e., typing up an item on a keyboard, phone, or tablet.

Some of the more innovative technologies engage consumers as they go about their ordinary shopping journey, encouraging more tactile interaction with product. By providing brightly lit shelves, video/animation, or interactive touch screens, smart fixtures grab consumers’ attention, encourage a “handshake” with the product, an activity that likely increases the consumer's propensity to buy, and provides product information, tips, and other opportunities for brand engagement.

At the same time, these fixtures provide some of the same sensing capabilities their less conspicuous counterparts give retailers, enabling retail firms to track which products attracted consumer to touch/feel/try them, providing basic demographic feedback, and offering emotion recognition to assess the nature of consumer reactions.

Smart dressing rooms Smart dressing rooms, engaging smart mirrors, use RFID technology to identify an item brought into a dressing room.

Oak Labs’ version, for example, which we sampled in Rebecca Minkoff’s Soho store, enables consumers to choose their language, recommend complementary items, and empowers the customer to engage store-level associates to get a different size or style without leaving the dressing room (relieving consumers of the burden of getting dressed/undressed simply to widen the range of goods to try on).

The key to this practice is RFID, with tagged inventory picked up by readers in the dressing room, allowing retailers to track what product is entering the fitting room and the speed at which items are converting from try-on to purchase. When a customer brings an RFID- tagged garment into the fitting room, interactive displays will recognize the items and will pull up product information and availability. From there, a customer can see other available sizes and colors in addition to availability online or in nearby stores. If she wants another size or color in-store, she can request a new size or color to try on in that store, or she can order the item online and have it shipped home – all of which can be accomplished from the fitting room. Some smart mirrors are already capable of processing payments through mobile payment platforms such as Apple Pay. Not only do these capabilities free up time for sales associates, but they also lead to more dynamic fulfillment, as the customer’s purchase options extend beyond the physical store.

Goldman Sachs Global Investment Research 45 August 2, 2017 Profiles in Innovation

The next steps for fitting room RFID will incorporate machine learning to provide an even more personalized experience for the customer. Eventually, bringing a garment into the fitting room will trigger the interactive display to show additional recommendations based on a customer’s purchase history and historical preferences. Tech firms are also developing “smart RFID” that can detect how long each garment was tried on for so that retailers can get a better sense of customer intent, which can be incorporated into future marketing strategies.

This concept is the classic example of an innovation than can have applications for both business intelligence and marketing. It enhances the consumer’s in-store experience, while passing along to the retailer or brand data on what consumers are trying on – which they can then track to actual sales, assessing conversion – how much time consumers spend in fitting rooms, and what proportion are not native English speakers.

Perch, an emerging startup in this space, emanated from Potion, a design-oriented business, and leads with its aesthetic appeal, while providing its retail enterprise customers a full content curation, and analytics via the cloud.

Digital signage Beabloo, based in , works to enhance proximity marketing in a retail context. It can enable a retailer to customize messaging on digital signage based on its best sense of who is walking by – inside or outside the store – based on age, gender, and engagement level, rendering displayed content more relevant, constantly testing its decision tree against sales outcomes.

The company leverages a combination of video feeds and beacon technologies to capture data about customers. Broader-lens, ceiling-mounted video sensors are used to track traffic trends across the store. Video feeds are also used in the actual digital signage with a small webcam at the top to determine exactly who is looking at the screen, which is then used to alter the content for the demographic of the shopper. The video feed is even able to capture the emotional response of a consumer based on the content of the digital signage for further strategic refining. The company has leveraged this functionality to help retailers test marketing campaigns, where a single marketing campaign is displayed on the screen that does not change based on the viewer, but the emotional response and level of engagement are used to help decide which campaigns to use more broadly in static- display scenarios. Like others cited in this report, Beabloo deploys beacons to track customer movement near and within the store but measures conversion vs. display signage to assess its efficacy.

Goldman Sachs Global Investment Research 46 August 2, 2017 Profiles in Innovation

Farfetch

The opportunity There are luxury brands with global appeal that lack the resources to develop an optimal ecommerce presence or even ecommerce capabilities. They are reluctant to reach out via sites that traffic in mass market brands, even via associated marketplaces, but keen to tap into audiences that transcend their boutique or wholesale presence. And, equipping their own retail stores with state of the art marketing and BI technology involves costs they may not be willing to bear, and human capital they may not possess.

The solution Farfetch, a private company based in the UK, addresses all of these opportunities. According to the UK magazine The Business of Fashion, it generated $800 million of GMV in 2016.

 Under the Farfetch name, it operates an online platform for high-end brands, giving a global base of consumer access to a curated array of fashionable brands in men’s, women’s, and kids’ apparel, overseeing delivery as well – including the capacity for same-day delivery in nine different fashion-oriented cities.

 Under the Black & White trade name, it provides white-label ecommerce solutions for individual brands.

 Most germane to our work, it recently introduced the contours of a Store-of-the-Future in London. Despite its core focus online, the company notes that luxury apparel actually under-indexes online, and is looking to make the most of retail traffic. Part of the premise is the consumer checking into the store, using their app to enable an interactive, personalized experience. Deploying RFID, a “magic rail” (think rack) detects when items are removed and populates them on the consumer’s “wish list,” and salient details of items placed on a magic rail in the dressing room can be projected on a “magic mirror,” enabling the customer to order different sizes and colors. Payment can be executed via mobile app as well. The data is retained to help the staff cater to the customer upon their next visit – and to sell in to that customer with an enhanced sense of their tastes. All of these data can be processed at a macro level as well, assessing which designers’ collections are attracting consumers, and which are not. The company intends to roll this out at acquired stores in London and New York in the near-term, and to license the technology over time.

The challenge We view concerns over privacy, and the awareness of data sharing, as the biggest impediment to a collaborative in- store data sharing experience for consumers and retailers.

Goldman Sachs Global Investment Research 47 August 2, 2017 Profiles in Innovation

Interview with Healey Cypher

Healey Cypher is the Co-Founder and CEO of Oak Labs, a retail tech startup dedicated to developing an interactive, technology-equipped retail experience. Prior to founding Oak Labs in 2015, Healey served as the Chief of Staff of Global Product Management and the Head of Retail Innovation at eBay.

Matthew Fassler: What led you and your colleagues to start the associate would bring it to you as fast as they possibly could, Oak? letting you stay in the fitting room not having to go wander in your socks trying to find someone. Healey Cypher: The starting point for me was about 8 years ago. eBay bought our startup and I became Chief of Staff to the CTO of ------eBay, and this was when it was eBay and PayPal and GSI. We did a lot of business development work with retailers, and as these “I don’t know if the word offline even exists retailers would come in we’d pitch them on e-comm solutions, and anymore, especially because of connected toward the ending they’d all say the exact same thing, this is great, now what solutions do you have for my physical stores, which are devices like mobile.” 95% of my sales? We just didn’t have a lot to say. I convinced the c------suite to let me start a retail innovation team, it was a 17-person, off- campus secret lab of sorts, and the whole thesis was: how do we When we build things like this, it’s not just kind of a random thesis, bring the best of online thinking into the physical world? We had we have a pretty specific methodology which is we think of in-store launches like kid’s pick Saturday window shops and Toys R Us gift experiences as identical to online experiences in that it is a finders, and we built the Rebecca Minkoff store down Green Street. conversion funnel. If you look at the top customers’ journeys weighted by volume and value, like e-comm starts to reduce the drop MF: Let’s talk a bit about how you’re carrying that mission out, off points between a home screen and a product page or product and about the contours of your core business today. page to cart or cart to checkout, you can start to do that in stores. If HC: We started doing all these things at eBay because we realized someone goes into a fitting room there’s a 2/3 chance you’re going to it’s actually pretty tough to do. It’s hard to build interactive convert, so we started to focus there to prove the ROI case in some experiences that are meaningful in stores, it’s hard to capture of these innovative technologies, and shown some incredible results. meaningful data. It’s hard to act upon that data and so what we MF: What do consumers think is missing from their in-store spent the last two years doing is building out the first modern experience, particularly as they now have more options operating system for interactive experiences in stores. It means you available to them online? could rapidly build beautiful new experiences on unexpected new form factors that solve real consumer needs, and part of this is HC: It’s the exact right question. Consumers are being conditioned driven by a market demand which is look, the ways that retail used every day by mobile devices. I just read a recent study that said that to compete which is on the three orthogonals of proximity, selection the average person uses a mobile phone 150 times a day. It is and price. Big stores, a lot of stuff, great deals; you can’t compete on training us every moment of every day to think things are personal those anymore. Today you’re competing on three different and fast and beautiful and robust and easy. When you walk into a orthogonals, it’s experience, it’s service, it’s urgency, and so store and you don’t have that experience, well, that’s a challenge; knowing that there is this conversion to physical and digital coming that’s kind of where they say well, why don’t I know if it’s in stock, as a result of this changing consumer demand accelerating and why isn’t the answer immediately available, why doesn’t it know who outpacing what retail can do, we built this platform because we also I am and then give me personal results? I think part of the challenge found that engineers were spending a lot of time building up the core of the kind of physical world – look I don’t know if the word offline versus the experience. even exists anymore, especially because of connected devices like mobile. Part of the challenge is how do you create that software An example of an experience we built just to make it really specific is defined future of retail where the store is changing around you. working with Ralph Lauren to launch interactive fitting rooms. Someone would walk into the fitting room on 5th Avenue or in Dallas MF: Now what about the problems that retailers are coming to and the mirror would actually come alive. It would read all the items you to solve? using radio frequency in the fitting room and then that mirror would HC: At the highest level, I think that retailers are very data stark. become interactive. It would say hey, here’s the things you have in There’s a myth out there in the world that online has better data than the fitting room, here’s other colors, sizes and cuts that are available the physical world and it is that, it’s a myth. If you instrument your right now. Here’s recommended items we think you might like based stores the right way the passive data collection, companies like on what you’re trying on right now. You can change lighting, you can RetailNext which have sensors, they’re getting a sense of heat maps change language and you can request other sizes and cuts and and then the active data collection, when you provide tools to

Goldman Sachs Global Investment Research 48 August 2, 2017 Profiles in Innovation

associate some tools with customers, is truly insane. I mean we that mobile point of sale fails in stores isn’t the app, it’s not the know things that fitting room product that no one has ever known gateway processing, it’s because the access points, the wi-fi before. Of the five sweaters in the Fall line, maybe three of them [connections], are slow. don’t convert at the same average conversion rate as the other MF: Could you just give us your perspective on that RFID saga SKUs in the store and we know customers are asking for a bigger within retail over the past 10 or 20 years? size. We know they’re asking for a different shade. That is actionable data, you can go back to your vendors, go back to merch, go back to HC: RFID is not new, it’s just become much cheaper. You know like the buy team, go back to the plot design team and make intelligent 10-12, 15 years ago Walmart started hitting RFID hard, to destroy decisions about inventory and stock. That’s one of them, just data supply chain metrics, and everyone was very excited about it. There starvation. I think the second is how do you bring intelligence in the are a lot of open questions about how you get the price down, and stores so it optimizes you selling more things at full price. Retail that what’s wonderful is now an RFID tag itself, like a beautiful one, is has been competing in the mid-tier which is trying to compete on four and a half cents and the net cost per garment if you do it at discounts, they have no more margin left to give up, so how do you scale including all the out backs is five and a half cents. If you roll optimize how many items you’re selling at full price? You surround it that up, I think between 20%-30% of all fashion apparel now has by a great experience, when you anticipate correctly, when you’re RFID tagging; it’s growing very quickly. not just kind of guessing and so things like inventory intelligence – by I was talking to the number one producer of RFID chips in the world, the way, industry average inventory intelligence is only 65% – one Impinj based out of Seattle, at their conference. They were saying life out of three things in the store, they don’t know if it’s in their store. to date we’re so proud to say that we have made ten billion chips by How do you help out with things like that? that date and this is over, I think it was like a seven-year period. ------Then, all of the sudden in the last year they came out and said they had made four billion in the last twelve months. That accelerated “I think as we move into a world where growth is insane. Look, RFID had some operational challenges. You shopping becomes more about the check-in have to figure out where you’re going to tag it, is it the manufacturer, is it the DC, is it the store? How do you do re-tags? But really what it than the check-out, think Uber, think Amazon is, to me, it’s about people in IT, people in retail realizing that you Go…[which means] people are going to have to invest in foundational technology to see the ROI, or, even a assume you’re collecting certain amounts of new term that I recently picked up which is ROO, the return on objective. When you create those enabling kinds of technologies you data, they’ve already preapproved ” can do things like dynamic fulfillment, like anticipating needs.

------MF: How does innovation help break down barriers between channels? The third thing is how do you make associates more impactful? The cost of employing people is going [in] a pretty specific direction, and HC: We can’t do anything successfully if they don’t have both the that’s up. As you think about the activities you’re asking your retail store ops team bought in and the e-comm team, the digital associates to take, how do you make them more effective? How do team, because they’re pulling in rendered graphics from our mind, you make them more efficient? How do you remove the repetitive recommendations using e-comm. But then we’re also getting the meaningless tasks and increase the amount of clienteling you’re associates to adopt the technology and use it. The Oak Stock Room doing, the amount of expertise they have so people seek them out? is a really simple product. If you make a request in the front of the house to an associate, they pull out their mobile device, scan the MF: What are the key elements of the tech stack that retailers UPC, see where it’s available in the store, and if it’s in the back of the need to implement? house. Instead of jumping on a radio they just tap the screen, HC: You have to have product catalog parity or at least an request it and then there’s a big touch screen in the back of the understanding of it. If you are trying to create experiences in your house. A light goes on, a sound goes bling, and it shows that stores that are only sold in store and you don’t have digital assets of beautiful image of what has to go up front, it shows where it has to those or romance copy, it’s going to limit what you’re able to do. go and, most importantly, a clock starts counting down, so the Obviously, that leads to things like intelligent recommendations, like associates are like, “oh man I have to fill that!” visual merchandising intelligence. It is an example of us using tremendous digital assets for in-store The second is inventory intelligence, you know you put RFID in your success, but it also means now we have metrics on service levels, store, [for] which net cost of garment now is about five and a half and we can help teams compete with one another to get a sense of cents, it takes your inventory accuracy from 55% up to 99.5%. That’s how those conversion levels are correlated. It also now becomes a where you can compete. hub where you can fulfill dynamically, you can get in-store orders. In the past, that’s something that probably wouldn’t have been possible. The third is infrastructure. I think so often people have DSL lines in Different organizations, different P&Ls competing for the same sale; their store, they don’t have great power and so as a result they’re they saw it as cannibalization, and I think now you’re seeing these unable to provide experiences. The number one reason, by the way, new organizational goals that aren’t truly aligned.

Goldman Sachs Global Investment Research 49 August 2, 2017 Profiles in Innovation

Amazon Go

The opportunity The grocery category is associated with a high degree of difficulty for ecommerce, and has experienced limited share migration online relative to other sectors. has been in business for nine years and has not achieved the share and traction we've seen for Amazon in many other categories. While the company recently announced an agreement to acquire Whole Foods, its launch of Amazon Go preceded this decision.

The solution Amazon has introduced Amazon Go, on a test-basis, open only to associates in Seattle.

For context, Go fits as one of several retail concept/technology experiments from Amazon, and as part of its grocery/CPG mosaic (Fresh, Subscribe and Save, Dash Buttons, Pantry, etc). The company launched Amazon Fresh in Seattle in 2007, then waited roughly six years before opening its second market, Los Angeles, in 2013. We have a seen a similar pace to its roll-out of other physical retail concepts, such as its Kindle mall kiosks and campus bookstores. Amazon Fresh launched in six new metro areas in 2016, and service areas have expanded in ten states.

It is designed to offer a frictionless shopping experience. The company uses computer vision, deep learning algorithms, and sensor fusion to provide what Amazon is calling “Just Walk Out Technology.”As our Internet team wrote on December 6, "The store uses machine learning, computer vision, sensors and other technologies to allow customers to make purchases by logging in via a smartphone app upon store entry, select products from store shelves, then leave the store without a checkout line. The store sells ready to eat meals, staples, meal kits and other convenience items. Purchases are automatically charged to the individual's Amazon Prime account." According to press articles, patents filed by Amazon use an initial smartphone type and an array of cameras and sensors to track the customer and their movement through stores.

As an example, excerpted from an AMZN patent filing (from June 2014!), “an image capture device (e.g., a camera) may capture a series of images of a user's hand before the hand crosses a plane into the inventory location and also capture a series of images of the user’s hand after it exits the inventory location. Based on a comparison of the images, it can be determined whether the user picked an item from the inventory location or placed an item into the inventory location.” And, “In addition to cameras, other input devices, such as pressure sensors, infrared sensors, a scale, load cells, a volume displacement sensor, a light curtain, etc., may be utilized…”

The company is publicizing the effort via a YouTube video, certainly more so than it has its emerging small-format bookstore test, though it does have a history of publicizing emerging innovation around holiday time (i.e. as it did with drones).

With Go, Amazon appears to be beating retailers at their own game, developing the capability to reduce friction in the brick & mortar experience by eliminating the checkout process. While scanners sped that process up - and improved accuracy – self-checkout reduced labor, and line-busting mobile checkout technologies accelerate it further at moments of peak activity, all of these involve a bottleneck at the end of the transaction.

Competitively, this could pose as big a challenge for elements of the restaurant sector as for food retail. Amazon Go can be viewed as another entrant to a growing field of "grab and go" restaurant alternatives that offer better value and/or convenience versus traditional restaurants, such as prepared food at grocery; fast casual (particularly those with well-

(continues on next page)

Goldman Sachs Global Investment Research 50 August 2, 2017 Profiles in Innovation

Amazon Go (continued)

executed digital ordering channels) as an alternative to casual; and C-stores. Similar to C-stores it could also come with a built in cost (and therefore price) advantage vs. restaurants as a result of a less labor-intensive model.

We see the greatest potential for this development to give AMZN an edge in businesses requiring little customer service, with lengthy transaction times relative to basket value, i.e. item-intensive transactions such as food. Moreover, the cultural dynamics associated with reducing store labor could be challenging to implement, particularly for firms with unionized workforces, though the emergence of technologies like self-checkout do not pose a competitive threat to the rest of the workforce (i.e. beyond the front-end). Also, the ability to track the consumer's every move would give AMZN a massively scaled trove of data on consumer behavior, offering vendors unparalleled insights on product placement and presentation.

Again, while it was not a part of AMZN’s portfolio at the time it developed the Go concept, the Whole Foods store base gives the company a massive lab for further testing, and a potential base for a rollout. Imagine a Whole Foods store with a mobile check-in option that substantially speeds the consumer’s shopping journey; that yields prompts consumers based on prior purchases as they walk by an aisle; that reminds consumers of unmet shopping needs based on comments to Alexa, hits to Dash buttons, or sensors in their kitchens.

The challenges The store was opened to associates in December with the goal of opening it to the public in early 2017. In that regard, it appears to be behind schedule. Based on our conversations with technology providers the cost of comprehensive sensor coverage, to the point of replacing all human interaction, is high. And, consumers could be concerned about privacy, though consumers willing to shop online - i.e. most consumers - would not be sharing any more about their shopping habits than they already do buying online.

The cost? Depends on scale Firms offering analytical solutions tend to operate on a SAAS model, with some cost for installation, and a monthly fee covering data capture, analysis, and recommendations. Firms we have engaged suggest the following model:

 50-100 sensors for full coverage – not full in the AmazonGo sense of full, but fairly comprehensive.

 10-15 sensors for partial coverage, tracking key metrics.

 A cost of $50-$100 per month per sensor.

 An installation fee of $500 per sensor.

Firms suggest that 10-15 stores can give retailers a relevant sample of consumer patterns for a chain of several hundred stores, with perhaps partial coverage at another 50-75. This yields an average annual cost of $2 million per firm and one-time installation of $500,000. That said, we see potential for far denser coverage to maximize both the analytical capabilities of a sensor network, and their role in clienteling. Doubling the penetration would suggest annual expenditures of $4 million per 500 stores, or $8,000 per store. Based on GS coverage of 110,000 stores, this suggests a total addressable market of $900 million per annum for our US coverage alone, with multiples of that once accounting for private companies and global players.

Goldman Sachs Global Investment Research 51 August 2, 2017 Profiles in Innovation

Who is investing in the store of the future?

Technologies we associate with retail innovation are attracting dramatically more venture capital than traditional retail, where venture investing has slowed to a trickle, though far less than pure-play ecommerce firms, according to CB Insights, as detailed below. For store-based retail innovation, we consider Mobile POS, POS hardware, retail and inventory software, and RFID systems.

Exhibit 13: VC funding in ecommerce and digital commerce enablement has significantly outpaced investment in store- based retail innovation and traditional retail VC funding; $ in millions

Ecommerce/ M- Retail (non- commerce Store-Based Retail Internet/mobile) Ecommerce Enablement1 Innovation2 2012 $50 $4,556 $281 $367 2013 55 5,157 343 227 2014 88 14,750 386 582 2015 72 23,540 478 495 2016 92 13,500 325 324 1Q17 1 4,300 102 107 2Q17 7 3,520 47 32 Total $364 $69,323 $1,962 $2,134

1 Ecommerce/M-commerce enablement includes deals in the follow CB Insights sub-industries: Ecommerce enablement and Mobile Commerce enablement 2 Store-Based Retail Innovation includes deals in the following CB Insights sub-industries: Mobile POS, POS hardware, Retail and inventory software (both in-store and online), RFID systems.

Source: CB Insights.

Of these categories, aside from ecommerce enablement, POS-oriented and inventory management firms have attracted the most venture funding. The largest deals are illustrated in Exhibit 15, the largest deals by sub-segment in Exhibit 16, and the largest investors (by deal count) in Exhibit 17.

Exhibit 14: VC activity for sub-industries in retail innovation since 2012

VC activity from Jan 2012 to present VC funding ($mn) Number of deals Ecommerce enablement $1,612 332 Mobile POS $816 69 Retail and inventory software (online) $565 132 Mobile commerce enablement $423 101 POS hardware $402 69 Retail/inventory software (in-store) $259 37 RFID systems $148 41 Total $4,225 781

Source: CB Insights.

Goldman Sachs Global Investment Research 52 August 2, 2017 Profiles in Innovation

Exhibit 15: Top 10 deals in retail innovation since 2012

Funding Amount Total Funding RETAIL INNOVATION Round Date ($mn) ($mn) Sector Industry Sub-industry Square Series D 17-Sep-12 200 717 Mobile & Telecommunications Mobile Software & Services Point of Sale RetailNext Series E 15-Apr-15 125 189 Software (non-internet/mobile) Retail & Inventory Software Shopify Series C 12-Dec-13 100 122 Internet eCommerce eCommerce enablement Coinbase Series C 20-Jan-15 75 117 Internet eCommerce eCommerce enablement Revel Systems Series C 11-Nov-14 65 129 Mobile & Telecommunications Mobile Software & Services Point of Sale Lightspeed POS Series C 16-Sep-15 61 126 Mobile & Telecommunications Mobile Software & Services Point of Sale ShopKeep POS Series D 28-Jul-15 60 97 Mobile & Telecommunications Mobile Software & Services Point of Sale NewStore Series B 18-Jul-17 50 97 Mobile & Telecommunications Mobile Commerce Mobile Commerce enablement Tillster Series F 26-Jun-13 50 105 Computer Hardware & Services Specialty Computer Hardware Point of Sales & Retail Computer Systems BigCommerce Series D 19-Nov-14 50 164 Internet eCommerce eCommerce enablement Source: CB Insights.

Goldman Sachs Global Investment Research 53 August 2, 2017 Profiles in Innovation

Exhibit 16: Top 10 deals by retail innovation sub-industry

Funding Amount Total Funding RETAIL AND INVENTORY SOFTWARE (ONLINE) Round Date ($mn) ($mn) Trax Image Recognition Series C 7-Jun-16 $40 $84 Toast Series B 5-Jan-16 $30 $37 I Believe Series B 30-Mar-17 $29 $29 Inturn Series B 2-Jun-17 $23 $36 Vend Series B 25-Mar-14 $20 $48 Feedvisor Series B 31-Jan-17 $20 $33 Stitch Labs Series B 20-Oct-15 $15 $24 First Insight Series C 5-Mar-15 $14 $22 Lengow Series B 2-Sep-15 $11 $13 Brightpearl Series D 19-Jan-16 $11 $34

Funding Amount Total Funding RETAIL AND INVENTORY SOFTWARE (IN-STORE) Round Date ($mn) ($mn) RetailNext Series E 15-Apr-15 $125 $189 Revionics Series D 8-Dec-14 $30 $59 Coherent Path Series A 24-Jun-14 $6 $7 Nexosis Seed VC - II 29-Nov-16 $5 $7 Weissbeerger Series A 1-Oct-15 $4 $4 Winnow Solutions Series A 20-Jan-16 $3 $4 Optii Solutions Series A 12-Apr-16 $3 $5 Ohana Companies Series A - III 25-Apr-13 $2 $17 Vcount Series A 14-Jul-16 $1 $1 Perch Interactive Seed VC 28-Apr-15 $1 $2

Funding Amount Total Funding MOBILE POINT-OF-SALE SOFTWARE Round Date ($mn) ($mn) Square Series D 17-Sep-12 $200 $717 Revel Systems Series C 11-Nov-14 $65 $129 Lightspeed POS Series C 16-Sep-15 $61 $126 ShopKeep POS Series D 28-Jul-15 $60 $97 Appetize Series A 20-Dec-16 $20 $20 CardFree Series A 10-Dec-12 $10 $14 LoopPay Series A 11-Dec-13 $10 $13 Gastrofix Series B - II 11-Jul-17 $8 $21 MokiMobility Series A 1-Oct-13 $7 $14 Talech Series A 27-Aug-14 $7 $7

Funding Amount Total Funding ECOMMERCE ENABLEMENT Round Date ($mn) ($mn) Shopify Series C 12-Dec-13 $100 $122 Coinbase Series C 20-Jan-15 $75 $117 MINDBODY Series E 20-Feb-14 $50 $113 BigCommerce Series D 19-Nov-14 $50 $164 Wikimart Series C 10-Sep-14 $40 $81 BeachMint Series D 25-Jan-12 $35 $74 Booker Software Series C 3-Mar-15 $35 $83 Wynd Series B 24-Nov-16 $32 $41 Tictail Series B 29-Jul-15 $22 $32 Symphony Commerce Series B 4-Sep-14 $22 $52

Funding Amount Total Funding MOBILE COMMERCE ENABLEMENT Round Date ($mn) ($mn) NewStore Series B 18-Jul-17 $50 $97 MoMo Series B 17-Mar-16 $28 $34 Curbside Series B 25-Jun-15 $25 $35 Shopgate Series C 25-Feb-16 $15 $22 Branding Brand Series B - II 26-Nov-13 $15 $33 Blispay Seed VC 31-Mar-16 $13 $26 PredictSpring Series A 28-Jun-16 $11 $11 Tocata Series A 21-Aug-12 $10 $10 Fril Series A 25-Sep-14 $9 $9 P97 Networks Series A - II 14-Jul-15 $9 $17

Funding Amount Total Funding POINT-OF-SALE HARDWARE Round Date ($mn) ($mn) Tillster Series F 26-Jun-13 $50 $105 E la Carte Series C 25-Sep-14 $35 $69 MedAvail Technologies Series C 17-Oct-14 $30 $30 Poynt Corporation Series B 22-Oct-15 $28 $28 KeyMe Series D 7-Dec-16 $25 $75 Miura Systems Unattributed VC 10-Jun-15 $16 $16 VendScreen Series B 10-Dec-12 $15 $31 Bingobox Series A 3-Jul-17 $14 $14 minuteKey Series B - II 3-Oct-12 $10 $59 Gastrofix Series B 2-Feb-17 $9 $13

Funding Amount Total Funding RFID SYSTEMS Round Date ($mn) ($mn) Omni-ID Series D 19-Jan-16 $21 $44 Altierre Series E 10-Jul-14 $21 $105 AwarePoint Series G 4-Jun-12 $14 $100 Famoco Series B 20-Feb-17 $12 $16 KNL Networks Series A 27-Oct-16 $10 $13 Biolog-id Series A 4-Nov-16 $8 $8 Resonant Angel 24-Jun-13 $7 $7 ASK Unattributed VC 10-Apr-12 $6 $6 Zest Labs Series B 21-Oct-12 $6 $77 TAMOCO Series A 21-Aug-14 $5 $5

Source: CB Insights.

Goldman Sachs Global Investment Research 54 August 2, 2017 Profiles in Innovation

Exhibit 17: Largest investors in ecommerce, digital enablement, and store-based retail innovation by number of deals

Ecommerce Since 2012 # of deals Last 2 years # of deals 500 Startups 156 500 Startups 70 Y Combinator 137 Y Combinator 65 500 Accelerator 118 500 Accelerator 53 Index Ventures 78 Techstars 33 Accel Partners 75 Rocket Internet 25 First Round Capital 72 FundersClub 24 Techstars 72 Greycroft Partners 24 Rocket Internet 65 Accel Partners 23 Tiger Global Management 65 Index Ventures 23 Andreessen Horowitz 62 Sequoia Capital India 22

Ecommerce/M-commerce enablement Since 2012 # of deals Last 2 years # of deals 500 Startups 20 Y Combinator 7 Y Combinator 13 FirstMark Capital 4 500 Accelerator 10 Startupbootcamp E-commerce 4 Bain Capital Ventures 9 General Catalyst 4 New York Angels 9 500 Accelerator 3 FirstMark Capital 8 500 Startups 3 Techstars 8 Bain Capital Ventures 3 Revolution 7 BPI 3 Balderton Capital 6 Kalaari Capital 3 East Ventures 6 Orange Digital Ventures 3 SV Angel 6 Ecommerce/M-commerce enablement includes deals in the follow CB Insights sub-industries: Ecommerce enablement and Mobile Commerce enablement

Store-based retail innovation Since 2012 # of deals Last 2 years # of deals Techstars 7 500 Accelerator 5 500 Accelerator 6 Square Peg Capital 4 Commerce Ventures 6 BPI France 3 Square Peg Capital 5 Food-X 3 Draper Athena 5 R/GA IoT Venture Studio 3 Food-X 5 Qualcomm Ventures 5 Y Combinator 5 yet2Ventures 5 Store-Based Retail Innovation includes deals in the following CB Insights sub-industries: Mobile POS, POS hardware, Retail and inventory software (both in-store and online), RFID systems.

Source: CB Insights.

Goldman Sachs Global Investment Research 55 August 2, 2017 Profiles in Innovation

Interview with …Activant’s Steve Sarracino

Steve Sarracino is the founder of Activant Capital, a growth equity firm. He focuses on innovative startups involved in commerce and in Internet of Things (IoT) technology, and has invested over $2 billion of equity across a broad range of technology companies since starting his career.

Matthew Fassler: Please describe the focus of your fund. and even Sales Forces – they just acquired Demandware. Also of those stacks that service retailers or CPG businesses are 20 years Steve Sarracino: We’re focused on growth stage businesses, and old. Demandware is almost 20 years old, the core of what Oracle has that bleeds in between venture capital and private equity, in two built in retail 20 years old, MICROS is a very old solution. Retailx, primary sectors. The first is commerce infrastructure, and we define SAP, same thing. You have retailers relying on these old, heavy ERP that as any time you are transacting a good or a service, the solutions where they inventory data and customer data siloed and underlying enterprise technology that enables it. It could be sales very difficult to reach, now they’re trying to move to this new enablement tools, fulfillment, back-end supply chain and logistics as omnichannel world where they’re going to have to compete with an well. The other area is full stack enterprise IT, taking hardware and Amazon, and the data needs to be real time. software and grabbing analog data in the real world and making it digital, and then using that for things like predictive analytics or ------sensor-driven decisioning. “…the core of the business has changed in MF: How would you describe the state of the venture capital and growth equity businesses today? that what used to be: a small seed round and

SS: There’s been an influx of capital through our sectors, in venture, then we’d move into venture A, B and C. That growth and PE, and primarily driven by low interest rates and people whole world has been flipped and people are chasing returns. The result has been valuations have been generally using much more money earlier just because up across the board. But the core of the business has changed from that what used to be a small seed round and then we’d move into there’s not only demand for investment venture A, B and C; that whole world has been flipped, and people opportunities but the amount of capital it are raising much more money earlier, because there’s not only demand for investment opportunities, but the amount of capital it takes to build a differentiated business.” takes to build a differentiated business. ------

The iPhone only came out 10 years ago, AWS is probably 5-6 years The experience for their customer needs to change, and that old at this point. To get a company up and running has been a lot provides this tremendous opportunity for us and look the stores; cheaper. What you saw was a massive amount of company creation, they’re not going away but they’re going to change dramatically, and apps, very simple tools that you could use at the enterprise level to it’s not just the way in which we shop and the look and feel of the drive engagement, both revenue increases and cost reductions. store but who owns them. When things burn in retail new things What we’re seeing now is that the quick wins have largely been blossom in their place, and so we’re seeing that and that provides taken. To build a large enterprise business requires much more tremendous opportunity for tech businesses. capital, and so we’re seeing some of the best firms raise more and more capital to allow those businesses to grow. It makes for MF: With a focus on the technology, what areas of retail demand opportunity because the competition has change in that venture the greatest focus from entrepreneurs today? What sectors and firms have to do more and more deals to hit the winners and the what business processes are attracting the most focus from growth firms can’t be wrong. It drives people to be focused in entrepreneurs, both those you’ve backed, and others? particular areas and that’s why we think, for our firm, being focused SS: The way we look at it, very simply, is they’re investing in any way on commerce and IOT provides differentiation. to service a customer better. So you have customer data, you have MF: Within the enablement sphere you have invested in some inventory data, and overlaying that you have the way that you’re businesses that serve the brick and mortar retail store. What interacting from a technological standpoint with the customer. In gave you the courage given the shift toward ecommerce? order to best serve your customer, what we’re finding is actually understanding their inventory, [how] you [can] get it to the customer, SS: You know as well as anyone the dislocation going on in retail, how you compete with Amazon with your set of inventory, and the and not just in retail but in the way that people buy goods and data that you have around your inventory. services, is higher than it’s ever been. When that happens there’s always opportunity for new technology and venture capitalists, if you Very large retailers don’t have 80-90% visibility in where their look at – look at the old line companies in tech, IBM, SAP, Oracle inventory lies, and if you want to transition your stores to be able to fulfill same day, to act like distribution centers, to have the right

Goldman Sachs Global Investment Research 56 August 2, 2017 Profiles in Innovation

inventory and the right placement someone is either in store, buying by the time they’re ready to acquire – and they acquire in a big way – online, you need to have perfect visibility into your supply chain. If that stack now is the old-line stack and we’re off investing in the you’ve ever seen a lot of capital being deployed behind providing newest one. Those organizations need to think about they can push better visibility to inventory, that’s the starting point and then that on tech and also do it in a way where they’re not loading the moves to really understanding your customer and omnichannel; company down with services, which was the old model. we’re seeing a lot of capital being deployed there. How do you know MF: How do your portfolio firms sell the economics to retailers? the same customer shopped online and then walked in the store and SS: All of these technology businesses try to sell ROI to the retailers, bought, what are the technologies that allow you to link that up? it’s generally very difficult to do, and the pricing model they’ve used is That’s where IOT comes in, because you can’t ignore what happens this SaaS – it’s just leasing software. The interesting thing is as in the analog world, you need to know when they’re in the store, who companies move to SAS they actually make themselves easier to be they are, what they bought – and batch updating with your POS ripped and replaced because you’re in the cloud, you’re not as sticky system at the end of the day is just not going to cut it anymore. and you have a fixed cost [for which] you have to really prove ROI. ------It’s not always clear where we think the world is going, and where our portfolio companies have moved is either a rev or cost share “The interesting thing is as companies move model and/or a transaction model. That is how you can prove to your to SAS they actually make themselves easier customer that you’re driving real value. When you’re taking some sort of percentage of the upside of revenue or cost savings. You’re at to be ripped and replaced because you’re in risk with the customer. There are horizontal SAS plays that won’t be the cloud, you’re not as sticky and you have a able to do this but in the vertical market I think this is where most of fixed cost that you have to really prove ROI.” the pricing is going to have to go. You have a beautiful, glossy product marketing scene, the ROI is 10x in a year. They’re not going ------to be able to take their word out, and they’re going to need these companies to take some risk with them, and so there’s more The second big area is understanding customer data. The key is to downside for the tech company but also more upside. overlay real mobile technology because mobile is the fastest growing part of online commerce by a factor of three. This is where Amazon MF: Three investments that you’ve made are central to the work has been successful, and if you look back to the old stacks, Oracle, that we’ve been doing; RetailNext, NewStore and Celect. Could SAP, IBM, they’re using old, heavy stacks that just don’t provide the you talk to each of them? real-time data to allow you to interact with your customers. The final SS: Retail Next has the most advanced sensor in the world, that has piece is getting the mobile OMS, the mobile Order Management deep learning, that basically you can do Amazon Go in your store. System, in place. So, it can track where people go, staff interaction, you can A-B test the floor of your store, just like you would your website, but it also Around the edges sure there’s lots of other interesting technologies can look to see, if you’re grabbing an apple off the shelf. And so, the but you have to go right to the core and redefine really how the end goal of Retail Next is to bridge that online/offline world. There’s retailer is operating. nothing like. It’s the only company that does [this] in the world; there MF: Have you seen some of the legacy leaders working to are others that do traffic counting, but this actually provides the upgrade their offer? overall picture of what’s happening in store.

SS: It’s easy to knock an Oracle or an IBM, – [but] the individuals NewStore is a mobile-first order management system, founded by there are very smart. We know a lot of them. They know where they Stephan Schambach, who founded Intershop and took it public and need to go, the challenge is that they’re hampered with these legacy he founded Demandware and took it public and sold it to Salesforce. solutions and the product managers and the product teams and the This is his third go-round and he is giving Amazon-like functionality to inertia around those legacy solutions are very powerful. They need brands and retailers, in a mobile format, then extending all the way to get to a place where they can cannibalize their own business in back to inventory warehouse management supply chain and then all order to provide what retailers and CPG companies need now and the way forward to CRM and clienteling in store. There’s actually for the next 10 to 15 years. The individuals inside these apps or progressive mobile web that the sales team in store can use organizations, they actually really get it and they would think about to work with their customers, and understand where inventory is. technology the same way we would around inventory, customer and And then our final company, Celect, is inventory management mobile but getting them there is going to take some time. software, developed by MIT professors. They built—so, it could be How are they going to get there? They’re probably going to end up called AI, our belief is AI doesn’t really exist, and we may talk about that but it’s extremely complex machine learning that takes limited having to acquire versus re-write their stack so you take like a amounts of data and is able to expand the data set and then help Demandware for instance that Salesforce bought, brilliant buy by you allocate inventory across all channels: online, stores, DCs, and Salesforce, great people, but that stack is not built for mobile. To do help increase inventory turns. So, this solves that, sort of, level one mobile properly you’re going to have to re-write the whole thing, and problem around inventory and how to manage your inventory across so it poses really interesting challenges, and what happens often is chain.

Goldman Sachs Global Investment Research 57 August 2, 2017 Profiles in Innovation

What could derail the store of the future?

Ecommerce displaces the physical store Improvement in ecommerce technology could replicate or replace some of the most critical functions served by a physical store. Ironically, most of these improvements – in areas such as conversational commerce and AR/VR – can enhance the in-store experience, but we believe they drive greater improvement from legacy practices to the optimized experience for online retail than for brick & mortar, and as such a superior in-store experience is at risk of getting eclipsed by an even more improved at-home experience.

AR imagery that can envision a consumer in a given outfit can be deployed equally well at home as in-store, so long as consumers own the hardware. Today, early in the roll-out of VR technology, with price points remaining high, stores have an edge, but that edge is unlikely to persist as prices come down and hardware becomes more ubiquitous.

Also, advancements in voice-based or “conversational” commerce could enhance the at- home shopping experience at the expense of the store.

The speed of retail store share loss to ecommerce could undermine innovation in three ways:

 Rapid volume declines could quickly undermine the economics of store-based retailers, rendering the notion of investment moot, especially if the costs of emerging technology do begin to not decline more rapidly. More retailers are declaring bankruptcy and liquidating, and in a mall-based environment, business failures are contagious, given the interdependency of tenants, and the need for traffic to maintain mall vitality.

 Retailers may choose to focus on their online-only offerings. To the extent that many retailers are most unprepared to engage in pure-play ecommerce, even if their stores are leaving much money on the table, they need to contend with the reality of customer loss today. As such, some view this investment path as a mere bandage that will barely address a more acute wound.

 The speed of innovation outpaces the pace of execution on current investment. Retailers need to scale investment across distributed store infrastructure, an effort that requires much time and money. Even if consumers are willing to maintain sufficient in-store spend to support big dollar investment in store-based IT, the nature of the consumer’s interaction with a retailer is moving quickly, which threatens to render big investments today less effective over time.

Who’s the sponsor? More so than in the past, the kind of in-store IT innovation we are discussing is the province of marketing and merchandising executives, more so than finance, supply chain, and traditional MIS managers. These executives are in unfamiliar territory lobbying for IT investment, such that these initiatives have to rise to the top of the C-suite to spur decisions. These factors have likely converged to drive a reality of paralysis, despite management’s best intentions.

For example, in the aftermath of a challenging holiday season, earlier this year Target stopped development on its “Store of the Future,” deploying robots for picking goods, and more space reserved for more experiential efforts such as classes, according to Recode and

Goldman Sachs Global Investment Research 58 August 2, 2017 Profiles in Innovation

later confirmed by the company, with CEO Brian Cornell telling Fortune magazine (March 20, 2017) that “he wanted to make sure ‘all of our innovation dollars are providing benefits in the near term.’”

Choppy implementation Entrepreneurs developing technology solutions for retailers tell us that rollouts sometimes fail because WiFi coverage is insufficient, and rollouts sometimes are hindered by insufficient electrical capacity (or outlets) in a box. Associates might be loathe to use technology that is clunky or cumbersome for them.

Privacy Retailers gathering data via sensors will need to self-regulate to avoid intrusive data capture, and all the more so when it comes to sharing such data. Moreover, consumers may prove reluctant to “opt-in” to sharing data, via apps or other means, in ways that would optimize retailers’ ability to cater to them.

Goldman Sachs Global Investment Research 59 August 2, 2017 Profiles in Innovation

Appendix

Exhibit 18: Capex breakout for covered retailers that disclose capex spend by category Capex disclosures for select retailers in FY2016

% of total % of total AEO Capex ($mn) Capex BBY Capex ($mn) Capex New stores, remodels and refurbishes $74 46% E-commerce and Information Technology $371 64% IT and distribution center improvements $47 29% Store-related projects (remodels and merchandising) $208 36% E-commerce $29 18% New stores $3 1% Home office projects $12 7% Other $0 0% Total Capex $162 Total Capex $582

BURL JCP New stores, store refreshes and remodels $85 55% Store renewals and updates $240 56% IT and other business initiatives $32 21% Capitalized software $100 23% Supply chain initiatives $22 14% Technology and other $70 16% Localization efforts $16 10% New and relocated stores $17 4% Total Capex $155 Total Capex $427

JWN KSS New stores, relocations and remodels $516 61% Information technology $353 46% Information technology $237 28% Store strategies $215 28% Other $93 11% Base capital $200 26% Total Capex $846 Total Capex $768

LB MIK Stores $772 78% Stores $70 61% Technology and infrastructure to support growth $218 22% Information systems $28 24% Total Capex $990 Corporate and other $17 15% Total Capex $114

ROST TGT Stores $208 70% Exisiting store investments $830 54% Information systems, corporate, and other $49 16% IT, supply chain, and other $587 38% Distribution and transportation $41 14% New stores $130 8% Total Capex $298 Total Capex $1,547

TJX TSCO Office and distribution centers $559 55% Stores $164 73% Store renovations and improvements $275 27% IT and distribution center improvements $62 27% New stores $191 19% Corporate and other $0 0% Total Capex $1,025 Total Capex $226

ULTA WMT New stores, including remodels and relocations $154 41% IT, distribution, digital retail, and other $4,162 39% Merchandising $83 22% New stores and clubs $2,171 20% Information Technology $56 15% Remodels $1,589 15% Supply Chain $41 11% Total US Capex $7,922 75% Store maintenance & Other $40 11% Walmart International $2,697 25% Total Capex $374 Total Capex $10,619

Source: Company data, Goldman Sachs Global Investment Research.

Goldman Sachs Global Investment Research 60 August 2, 2017 Profiles in Innovation

Financial Advisory Disclosure Goldman Sachs and/or one of its affiliates is acting as a financial advisor in connection with an announced strategic matter involving the following company or one of its affiliates: Macy's, Inc., , Inc., Amazon.com, Inc.

Goldman Sachs Global Investment Research 61 August 2, 2017 Profiles in Innovation

Disclosure Appendix Reg AC We, Matthew J. Fassler, Jesse Hulsing, Heath P. Terry, CFA, Lindsay Drucker Mann, CFA, Katie Price, Rachel Binder, Heather Bellini, CFA, Matthew Cabral, Natasha de la Grense, Richard Edwards, Hugo Scott-Gall, Toshiya Hari, Rob Joyce, Chandni Luthra, Piyush Mubayi, James Schneider, Ph.D., Stephen Tanal, CFA and Alexandra Walvis, CFA, hereby certify that all of the views expressed in this report accurately reflect our personal views about the subject company or companies and its or their securities. We also certify that no part of our compensation was, is or will be, directly or indirectly, related to the specific recommendations or views expressed in this report.

Unless otherwise stated, the individuals listed on the cover page of this report are analysts in Goldman Sachs' Global Investment Research division. GS Factor Profile The Goldman Sachs Factor Profile provides investment context for a stock by comparing key attributes to the market (i.e. our coverage universe) and its sector peers. The four key attributes depicted are: Growth, Financial Returns, Multiple (e.g. valuation) and Integrated (a composite of Growth, Financial Returns and Multiple). Growth, Financial Returns and Multiple are calculated by using normalized ranks for specific metrics for each stock. The normalized ranks for the metrics are then averaged and converted into percentiles for the relevant attribute. The precise calculation of each metric may vary depending on the fiscal year, industry and region, but the standard approach is as follows: Growth is based on a stock's forward-looking sales growth, EBITDA growth and EPS growth (for financial stocks, only EPS and sales growth), with a higher percentile indicating a higher growth company. Financial Returns is based on a stock's forward-looking ROE, ROCE and CROCI (for financial stocks, only ROE), with a higher percentile indicating a company with higher financial returns. Multiple is based on a stock's forward-looking P/E, P/B, price/dividend (P/D), EV/EBITDA, EV/FCF and EV/Debt Adjusted Cash Flow (DACF) (for financial stocks, only P/E, P/B and P/D), with a higher percentile indicating a stock trading at a higher multiple. The Integrated percentile is calculated as the average of the Growth percentile, Financial Returns percentile and (100% - Multiple percentile). Financial Returns and Multiple use the Goldman Sachs analyst forecasts at the fiscal year-end at least three quarters in the future. Growth uses inputs for the fiscal year at least seven quarters in the future compared with the year at least three quarters in the future (on a per-share basis for all metrics). For a more detailed description of how we calculate the GS Factor Profile, please contact your GS representative. Quantum Quantum is Goldman Sachs' proprietary database providing access to detailed financial statement histories, forecasts and ratios. It can be used for in-depth analysis of a single company, or to make comparisons between companies in different sectors and markets. GS SUSTAIN GS SUSTAIN is a global investment strategy aimed at long-term, long-only performance with a low turnover of ideas. The GS SUSTAIN focus list includes leaders our analysis shows to be well positioned to deliver long term outperformance through sustained competitive advantage and superior returns on capital relative to their global industry peers. Leaders are identified based on quantifiable analysis of three aspects of corporate performance: cash return on cash invested, industry positioning and management quality (the effectiveness of companies' management of the environmental, social and governance issues facing their industry). Disclosures Coverage group(s) of stocks by primary analyst(s) Matthew J. Fassler: America-Retail: Specialty Hardlines. Jesse Hulsing: America-Emerging Software. Heath P. Terry, CFA: America-Internet. Lindsay Drucker Mann, CFA: America-Accessories: Handbags and Jewelry, America-Brands: Athletic and Other Wholesale Brands, America-Retail: Department Stores & Specialty, America-Retail: Off-Price. Heather Bellini, CFA: America-Software. Matthew Cabral: America-IT Hardware. Richard Edwards: Europe-General Retail. Toshiya Hari: America-Semiconductor Capital Equipment, America-Semiconductors. Rob Joyce: Europe-Food Retail, Europe-General Retail. Piyush Mubayi: China Internet. James Schneider, Ph.D.: America-ATM/POS and Self-Service, America-IT Consulting and Outsourcing, America-Transaction Processors. Stephen Tanal, CFA: America-Specialty Hardlines, America-Supermarkets, Dollar Stores, & Convenience. America-ATM/POS and Self-Service: CPI Card Group, CPI Card Group, VeriFone Systems Inc.. America-Accessories: Handbags and Jewelry: Coach Inc., Michael Kors Holdings, Signet Jewelers Ltd., Tiffany & Co.. America-Brands: Athletic and Other Wholesale Brands: Canada Goose Holdings, Canada Goose Holdings, Columbia Sportswear Co., Hanesbrands Inc., Nike Inc., PVH Corp., Ralph Lauren Corp., Under Armour Inc., VF Corp.. America-Emerging Software: Alteryx Inc., Appian Corp., Apptio Inc., Benefitfocus Inc., Blackline Inc., Cornerstone OnDemand Inc., Guidewire Software Inc., Hortonworks Inc., HubSpot Inc., Intuit Inc., MuleSoft Inc., New Relic Inc., ServiceNow Inc., Shopify Inc., Splunk Inc., Tableau Software, Talend SA, Teradata Corp., Ultimate Software Group, Veeva Systems Inc., Zendesk Inc.. America-IT Consulting and Outsourcing: Accenture Plc, Black Knight Financial Services Inc., CGI Group, CGI Group, Cognizant Technology Solutions, DXC Technology Co., Fidelity National Information Services, Fiserv Inc., International Business Machines, Sabre Corp., West Corp.. America-IT Hardware: 3D Systems Corp., Aerohive Networks Inc., CDW Corp., Motorola Solutions Inc., Presidio Inc., Stratasys Ltd., Xerox Corp., Zebra Technologies Corp.. America-Internet: Amazon.com Inc., Bankrate Inc., Blue Apron Holdings, Criteo SA, eBay Inc., Endurance International Group, Etsy Inc., Expedia Inc., Groupon Inc., GrubHub Inc., IAC/InterActiveCorp, LendingClub Corp., Match Group, Netflix Inc., Pandora Media Inc., PayPal Holdings, Priceline.com Inc., Shutterfly Inc., Snap Inc., TripAdvisor Inc., Trivago N.V., TrueCar, Twitter Inc., WebMD Health Corp., Yelp Inc., Zillow Group, Zynga Inc.. America-Retail: Department Stores & Specialty: American Eagle Outfitters Inc., Gap Inc., Kohl's Corp., L Brands Inc., lululemon athletica inc., Macy's Inc., Nordstrom Inc., Urban Outfitters Inc.. America-Retail: Off-Price: Burlington Stores Inc., Ross Stores Inc., TJX Cos.. America-Retail: Specialty Hardlines: Advance Auto Parts Inc., At Home Group, AutoZone Inc., Bed Bath & Beyond Inc., Best Buy Co., CarMax Inc., Container Store Group, Costco Wholesale, Floor & Decor Holdings, Genuine Parts Co., Home Depot Inc., KAR Auction Services Inc., Lowe's Cos.,

Goldman Sachs Global Investment Research 62 August 2, 2017 Profiles in Innovation

Lumber Liquidators Holdings, Michaels Cos., Monro Muffler Brake Inc., O'Reilly Automotive Inc., Office Depot, RH, Staples Inc., Target Corp., Ulta Beauty Inc., Wal-Mart Stores Inc., Wayfair Inc., Williams-Sonoma Inc.. America-Semiconductor Capital Equipment: Applied Materials Inc., Entegris Inc., Keysight Technologies Inc., KLA-Tencor Corp., Lam Research Corp., Teradyne Inc., Versum Materials Inc.. America-Semiconductors: Advanced Micro Devices Inc., Analog Devices Inc., Broadcom Ltd., Integrated Device Technology Inc., Intel Corp., Maxim Integrated Products, Nvidia Corp., NXP Semiconductors NV, Qorvo Inc., Skyworks Solutions Inc., Texas Instruments Inc., Xilinx Corp.. America-Software: Adobe Systems Inc., Akamai Technologies Inc., Alphabet Inc., Atlassian Corp., Autodesk Inc., Citrix Systems Inc., Facebook Inc., Microsoft Corp., MobileIron Inc., Okta Inc., Oracle Corp., Red Hat Inc., RingCentral, Salesforce.com Inc., Twilio, VMware Inc., Workday Inc.. America-Specialty Hardlines: Cabela's Inc., Dick's Sporting Goods, GNC Holdings, Hibbett Sports Inc., Party City Holdco Inc., Sportsman's Warehouse Holdings, Tractor Supply Co., Vitamin Shoppe Inc.. America-Supermarkets, Dollar Stores, & Convenience: Casey's General Stores Inc., Corp., Stores Inc., Inc., Kroger Co., Inc., SUPERVALU Inc., United Natural Foods Inc., Whole Foods Market Inc.. America-Transaction Processors: Automatic Data Processing Inc., Blackhawk Network Holdings, Evertec Inc., First Data Corp., FleetCor Technologies Inc., Global Payments Inc., MasterCard Inc., MoneyGram International Inc., Paychex Inc., Square Inc., Total System Services Inc., Vantiv Inc., Visa Inc., Western Union Co., WEX Inc.. China Internet: 58.com Inc., Alibaba Group, Baidu.com Inc., Ctrip.com International, Gridsum, JD.com Inc., NetEase Inc., New Oriental Education & Technology, SINA Corp., TAL Education Group, Tarena International Inc., Tencent Holdings, Vipshop Holdings, Weibo Corp.. Europe-Food Retail: NV, Booker Group, Carrefour, Casino, Colruyt, J Sainsbury, Jeronimo Martins, Metro, Morrison (Wm), Ocado Group, . Europe-General Retail: adidas, ASOS Plc, Associated British Foods, B&M European Value Retail SA, Burberry, Debenhams, Dixons Carphone Plc, Europris ASA, Hennes & Mauritz, Hugo Boss AG, Inditex, JUST EAT, Kering, Kingfisher, Luxottica (Italy), LVMH Moet-Hennessy Louis Vuitton, Maisons du Monde SAS, Marks & Spencer, Moncler SpA, Next, OVS SpA, Pandora, Pets at Home Group, Prada SpA, Puma, Richemont, Rocket Internet SE, Salvatore Ferragamo SpA, Showroomprivé, Sports Direct International Plc, Steinhoff International Holdings, Swatch Group, Takeaway.com, Technogym SpA, Ted Baker, Thule Group, Tod's, Tokmanni Group, XXL ASA, YOOX Net-A-Porter Group, Zalando SE. Company-specific regulatory disclosures Compendium report: please see disclosures at http://www.gs.com/research/hedge.html. Disclosures applicable to the companies included in this compendium can be found in the latest relevant published research Distribution of ratings/investment banking relationships Goldman Sachs Investment Research global Equity coverage universe

Rating Distribution Investment Banking Relationships Buy Hold Sell Buy Hold Sell Global 32% 54% 14% 65% 56% 49% As of July 1, 2017, Goldman Sachs Global Investment Research had investment ratings on 2,753 equity securities. Goldman Sachs assigns stocks as Buys and Sells on various regional Investment Lists; stocks not so assigned are deemed Neutral. Such assignments equate to Buy, Hold and Sell for the purposes of the above disclosure required by the FINRA Rules. See 'Ratings, Coverage groups and views and related definitions' below. The Investment Banking Relationships chart reflects the percentage of subject companies within each rating category for whom Goldman Sachs has provided investment banking services within the previous twelve months. Price target and rating history chart(s) Compendium report: please see disclosures at http://www.gs.com/research/hedge.html. Disclosures applicable to the companies included in this compendium can be found in the latest relevant published research Regulatory disclosures Disclosures required by United States laws and regulations See company-specific regulatory disclosures above for any of the following disclosures required as to companies referred to in this report: manager or co-manager in a pending transaction; 1% or other ownership; compensation for certain services; types of client relationships; managed/co- managed public offerings in prior periods; directorships; for equity securities, market making and/or specialist role. Goldman Sachs trades or may trade as a principal in debt securities (or in related derivatives) of issuers discussed in this report. The following are additional required disclosures: Ownership and material conflicts of interest: Goldman Sachs policy prohibits its analysts, professionals reporting to analysts and members of their households from owning securities of any company in the analyst's area of coverage. Analyst compensation: Analysts are paid in part based on the profitability of Goldman Sachs, which includes investment banking revenues. Analyst as officer or director: Goldman Sachs policy generally prohibits its analysts, persons reporting to analysts or members of their households from serving as an officer, director or advisor of any company in the analyst's area of coverage. Non-U.S. Analysts: Non-U.S. analysts may not be associated persons of Goldman, Sachs & Co. and therefore may not be subject to FINRA Rule 2241 or FINRA Rule 2242 restrictions on communications with subject company, public appearances and trading securities held by the analysts. Distribution of ratings: See the distribution of ratings disclosure above. Price chart: See the price chart, with changes of ratings and price targets in prior periods, above, or, if electronic format or if with respect to multiple companies which are the subject of this report, on the Goldman Sachs website at http://www.gs.com/research/hedge.html. Additional disclosures required under the laws and regulations of jurisdictions other than the United States The following disclosures are those required by the jurisdiction indicated, except to the extent already made above pursuant to United States laws and regulations. Australia: Goldman Sachs Australia Pty Ltd and its affiliates are not authorised deposit-taking institutions (as that term is defined in the Banking Act 1959 (Cth)) in Australia and do not provide banking services, nor carry on a banking business, in Australia. This research, and any access to it, is intended only for "wholesale clients" within the meaning of the Australian Corporations Act, unless otherwise agreed by Goldman Sachs. In producing research reports, members of the Global Investment Research Division of Goldman Sachs Australia may attend site visits and

Goldman Sachs Global Investment Research 63 August 2, 2017 Profiles in Innovation

other meetings hosted by the issuers the subject of its research reports. In some instances the costs of such site visits or meetings may be met in part or in whole by the issuers concerned if Goldman Sachs Australia considers it is appropriate and reasonable in the specific circumstances relating to the site visit or meeting. Brazil: Disclosure information in relation to CVM Instruction 483 is available at http://www.gs.com/worldwide/brazil/area/gir/index.html. Where applicable, the Brazil-registered analyst primarily responsible for the content of this research report, as defined in Article 16 of CVM Instruction 483, is the first author named at the beginning of this report, unless indicated otherwise at the end of the text. Canada: Goldman Sachs Canada Inc. is an affiliate of The Goldman Sachs Group Inc. and therefore is included in the company specific disclosures relating to Goldman Sachs (as defined above). Goldman Sachs Canada Inc. has approved of, and agreed to take responsibility for, this research report in Canada if and to the extent that Goldman Sachs Canada Inc. disseminates this research report to its clients. Hong Kong: Further information on the securities of covered companies referred to in this research may be obtained on request from Goldman Sachs (Asia) L.L.C. India: Further information on the subject company or companies referred to in this research may be obtained from Goldman Sachs (India) Securities Private Limited, Research Analyst - SEBI Registration Number INH000001493, 951-A, Rational House, Appasaheb Marathe Marg, Prabhadevi, Mumbai 400 025, India, Corporate Identity Number U74140MH2006FTC160634, Phone +91 22 6616 9000, Fax +91 22 6616 9001. Goldman Sachs may beneficially own 1% or more of the securities (as such term is defined in clause 2 (h) the Indian Securities Contracts (Regulation) Act, 1956) of the subject company or companies referred to in this research report. Japan: See below. Korea: Further information on the subject company or companies referred to in this research may be obtained from Goldman Sachs (Asia) L.L.C., Seoul Branch. New Zealand: Goldman Sachs New Zealand Limited and its affiliates are neither "registered banks" nor "deposit takers" (as defined in the Reserve Bank of New Zealand Act 1989) in New Zealand. This research, and any access to it, is intended for "wholesale clients" (as defined in the Financial Advisers Act 2008) unless otherwise agreed by Goldman Sachs. Russia: Research reports distributed in the Russian Federation are not advertising as defined in the Russian legislation, but are information and analysis not having product promotion as their main purpose and do not provide appraisal within the meaning of the Russian legislation on appraisal activity. Singapore: Further information on the covered companies referred to in this research may be obtained from Goldman Sachs (Singapore) Pte. (Company Number: 198602165W). Taiwan: This material is for reference only and must not be reprinted without permission. Investors should carefully consider their own investment risk. Investment results are the responsibility of the individual investor. : Persons who would be categorized as retail clients in the United Kingdom, as such term is defined in the rules of the Financial Conduct Authority, should read this research in conjunction with prior Goldman Sachs research on the covered companies referred to herein and should refer to the risk warnings that have been sent to them by Goldman Sachs International. A copy of these risks warnings, and a glossary of certain financial terms used in this report, are available from Goldman Sachs International on request. European Union: Disclosure information in relation to Article 4 (1) (d) and Article 6 (2) of the European Commission Directive 2003/125/EC is available at http://www.gs.com/disclosures/europeanpolicy.html which states the European Policy for Managing Conflicts of Interest in Connection with Investment Research. Japan: Goldman Sachs Japan Co., Ltd. is a Financial Instrument Dealer registered with the Kanto Financial Bureau under registration number Kinsho 69, and a member of Japan Securities Dealers Association, Financial Futures Association of Japan and Type II Financial Instruments Firms Association. Sales and purchase of equities are subject to commission pre-determined with clients plus consumption tax. See company-specific disclosures as to any applicable disclosures required by Japanese stock exchanges, the Japanese Securities Dealers Association or the Japanese Securities Finance Company. Ratings, coverage groups and views and related definitions Buy (B), Neutral (N), Sell (S) -Analysts recommend stocks as Buys or Sells for inclusion on various regional Investment Lists. Being assigned a Buy or Sell on an Investment List is determined by a stock's total return potential relative to its coverage. Any stock not assigned as a Buy or a Sell on an Investment List with an active rating (i.e., a stock that is not Rating Suspended, Not Rated, Coverage Suspended or Not Covered), is deemed Neutral. Each regional Investment Review Committee manages various regional Investment Lists to a global guideline of 25%-35% of stocks as Buy and 10%- 15% of stocks as Sell; however, the distribution of Buys and Sells in any particular analyst’s coverage group may vary as determined by the regional Investment Review Committee. Additionally, each Investment Review Committee manages Regional Conviction lists, which represent investment recommendations focused on the size of the total return potential and/or the likelihood of the realization of the return across their respective areas of coverage. The addition or removal of stocks from such Conviction lists do not represent a change in the analysts’ investment rating for such stocks. Total return potential represents the upside or downside differential between the current share price and the price target, including all paid or anticipated dividends, expected during the time horizon associated with the price target. Price targets are required for all covered stocks. The total return potential, price target and associated time horizon are stated in each report adding or reiterating an Investment List membership. Coverage groups and views: A list of all stocks in each coverage group is available by primary analyst, stock and coverage group at http://www.gs.com/research/hedge.html. The analyst assigns one of the following coverage views which represents the analyst's investment outlook on the coverage group relative to the group's historical fundamentals and/or valuation. Attractive (A). The investment outlook over the following 12 months is favorable relative to the coverage group's historical fundamentals and/or valuation. Neutral (N). The investment outlook over the following 12 months is neutral relative to the coverage group's historical fundamentals and/or valuation. Cautious (C). The investment outlook over the following 12 months is unfavorable relative to the coverage group's historical fundamentals and/or valuation. Not Rated (NR). The investment rating and target price have been removed pursuant to Goldman Sachs policy when Goldman Sachs is acting in an advisory capacity in a merger or strategic transaction involving this company and in certain other circumstances. Rating Suspended (RS). Goldman Sachs Research has suspended the investment rating and price target for this stock, because there is not a sufficient fundamental basis for determining, or there are legal, regulatory or policy constraints around publishing, an investment rating or target. The previous investment rating and price target, if any, are no longer in effect for this stock and should not be relied upon. Coverage Suspended (CS). Goldman Sachs has suspended coverage of this company. Not Covered (NC). Goldman Sachs does not cover this company. Not Available or Not Applicable (NA). The information is not available for display or is not applicable. Not Meaningful (NM). The information is not meaningful and is therefore excluded. Global product; distributing entities The Global Investment Research Division of Goldman Sachs produces and distributes research products for clients of Goldman Sachs on a global basis. Analysts based in Goldman Sachs offices around the world produce equity research on industries and companies, and research on macroeconomics, currencies, commodities and portfolio strategy. This research is disseminated in Australia by Goldman Sachs Australia Pty Ltd (ABN 21 006 797 897); in Brazil by Goldman Sachs do Brasil Corretora de Títulos e Valores Mobiliários S.A.; in Canada by either Goldman Sachs Canada Inc. or Goldman, Sachs & Co.; in Hong Kong by Goldman Sachs (Asia) L.L.C.; in India by Goldman Sachs (India) Securities Private Ltd.; in Japan by Goldman Sachs Japan Co., Ltd.; in the Republic of Korea by Goldman Sachs (Asia) L.L.C., Seoul Branch; in New Zealand by Goldman Sachs New Zealand Limited; in Russia by OOO Goldman Sachs; in Singapore by Goldman Sachs (Singapore) Pte. (Company Number: 198602165W); and in the United States of America by Goldman, Sachs & Co. Goldman Sachs International has approved this research in connection with its distribution in the United Kingdom and European Union. European Union: Goldman Sachs International authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority and the Prudential Regulation Authority, has approved this research in connection with its distribution in the European Union and United Kingdom;

Goldman Sachs Global Investment Research 64 August 2, 2017 Profiles in Innovation

Goldman Sachs AG and Goldman Sachs International Zweigniederlassung Frankfurt, regulated by the Bundesanstalt für Finanzdienstleistungsaufsicht, may also distribute research in . General disclosures This research is for our clients only. Other than disclosures relating to Goldman Sachs, this research is based on current public information that we consider reliable, but we do not represent it is accurate or complete, and it should not be relied on as such. The information, opinions, estimates and forecasts contained herein are as of the date hereof and are subject to change without prior notification. We seek to update our research as appropriate, but various regulations may prevent us from doing so. Other than certain industry reports published on a periodic basis, the large majority of reports are published at irregular intervals as appropriate in the analyst's judgment. Goldman Sachs conducts a global full-service, integrated investment banking, investment management, and brokerage business. We have investment banking and other business relationships with a substantial percentage of the companies covered by our Global Investment Research Division. Goldman, Sachs & Co., the United States broker dealer, is a member of SIPC (http://www.sipc.org). Our salespeople, traders, and other professionals may provide oral or written market commentary or trading strategies to our clients and principal trading desks that reflect opinions that are contrary to the opinions expressed in this research. Our asset management area, principal trading desks and investing businesses may make investment decisions that are inconsistent with the recommendations or views expressed in this research. The analysts named in this report may have from time to time discussed with our clients, including Goldman Sachs salespersons and traders, or may discuss in this report, trading strategies that reference catalysts or events that may have a near-term impact on the market price of the equity securities discussed in this report, which impact may be directionally counter to the analyst's published price target expectations for such stocks. Any such trading strategies are distinct from and do not affect the analyst's fundamental equity rating for such stocks, which rating reflects a stock's return potential relative to its coverage group as described herein. We and our affiliates, officers, directors, and employees, excluding equity and credit analysts, will from time to time have long or short positions in, act as principal in, and buy or sell, the securities or derivatives, if any, referred to in this research. The views attributed to third party presenters at Goldman Sachs arranged conferences, including individuals from other parts of Goldman Sachs, do not necessarily reflect those of Global Investment Research and are not an official view of Goldman Sachs. Any third party referenced herein, including any salespeople, traders and other professionals or members of their household, may have positions in the products mentioned that are inconsistent with the views expressed by analysts named in this report. This research is not an offer to sell or the solicitation of an offer to buy any security in any jurisdiction where such an offer or solicitation would be illegal. It does not constitute a personal recommendation or take into account the particular investment objectives, financial situations, or needs of individual clients. Clients should consider whether any advice or recommendation in this research is suitable for their particular circumstances and, if appropriate, seek professional advice, including tax advice. The price and value of investments referred to in this research and the income from them may fluctuate. Past performance is not a guide to future performance, future returns are not guaranteed, and a loss of original capital may occur. Fluctuations in exchange rates could have adverse effects on the value or price of, or income derived from, certain investments. Certain transactions, including those involving futures, options, and other derivatives, give rise to substantial risk and are not suitable for all investors. Investors should review current options disclosure documents which are available from Goldman Sachs sales representatives or at http://www.theocc.com/about/publications/character-risks.jsp. Transaction costs may be significant in option strategies calling for multiple purchase and sales of options such as spreads. Supporting documentation will be supplied upon request. All research reports are disseminated and available to all clients simultaneously through electronic publication to our internal client websites. Not all research content is redistributed to our clients or available to third-party aggregators, nor is Goldman Sachs responsible for the redistribution of our research by third party aggregators. For research, models or other data available on a particular security, please contact your sales representative or go to http://360.gs.com. Disclosure information is also available at http://www.gs.com/research/hedge.html or from Research Compliance, 200 West Street, New York, NY 10282. © 2017 Goldman Sachs. No part of this material may be (i) copied, photocopied or duplicated in any form by any means or (ii) redistributed without the prior written consent of The Goldman Sachs Group, Inc.

Goldman Sachs Global Investment Research 65 Mindcraft: Our Thematic Deep Dives Innovation & Disruption Music in the Air Factory of the Precision Advanced Virtual Reality Drones Future Blockchain Farming Materials Space Opportunity Risks

Insights & Policy Artificial China Rethinking Unfolding Healthcare’s Intelligence Batteries Mobility OLED Top of Mind Fortnightly US Infrastructure Thoughts Holy Grail Trade Policy

Commodity Corner The Low Carbon Economy

More Lean, The Great Battery Race Top Oil & Gas Projects Shale Innovation Copper Top Projects Promising Tech More Green

Rising Asia Quantamentals Consumer The Asia Stock MSCI Primer Apple Suppliers’ Dilemma Close-Up Millennials E+Commerce Collection

Consumer Currents GS SUSTAIN ...and more eCommerce’s The PM’s Guide to ESG Introducing GS SUSTAIN 50 The Rise of Craft Infinite Shelf