Retail AI Trends To Watch In 2021 2021 WHAT IS CB INSIGHTS? CB Insights helps the world’s leading companies make smarter technology decisions with data, not opinion. Our Technology Insights Platform provides companies with comprehensive data, expert insights and work management tools to drive growth and improve operations with technology. SIGN UP FOR A FREE TRIAL Table of Contents AI will become indispensable for 6 e-commerce fraud prevention Retailers will invest more in micro- 9 fulfillment infrastructure Supermarkets will tackle food waste 12 reduction with AI Retailers will double down on first-party 15 data strategy in a post-cookie world Auto tagging will become a must-have 18 AI tool for online retail AI will power hyper-local inventory 22 planning at the store level Checkout-free solutions will become 25 more accessible for retailers 3 Facing supply chain disruptions and dramatic changes in consumer behavior, artificial intelligence became central to how retailers adapted during the Covid-19 pandemic. As restrictions lift and stores reopen, here are the AI trends and technologies that will reshape the retail industry. When the Covid-19 pandemic hit, retail operations changed overnight. Brick-and-mortar locations closed. Customers shopped online at record levels, with knock-on effects in cybercrime and fraud. Supply chain disruptions led to product shortages and unpredictability. Forecasting models were rendered moot. AI for retail took a hit in 2020 as pandemic-related uncertainties shook investors’ confidence. Funding to the space fell 30% yearover-year in 2020, and deals reached a 4-year-low. Well-known vendors like autonomous store startup Stockwell AI (backed by GV, New Enterprise Associates, and others) closed shop. Retail AI Trends To Watch In 2021 4 But as new problems arose in the retail space, new AI solutions emerged. These tools helped brands adapt to the swift and unprecedented change, and they will continue to impact the industry long after the pandemic subsides. Retail AI has already hit a record funding level in 2021, driven by mega-rounds ($100M+) to vendors tackling issues like e-commerce fraud, e-commerce fulfillment, and first-party data analytics. In this report, we examine 7 retail AI trends that accelerated in 2020 and dig into what comes next for the space. 5 AI will become indispensable for e-commerce fraud prevention With a dramatic increase in online purchases and new customer authentication standards from the European Commission, e-commerce fraud prevention is top of mind for online merchants. To fight online fraud, retailers and brands require tools that protect user accounts and payment, identify illegitimate transactions, and prevent abuse of purchase and return policies — all while maintaining a seamless user experience. The pandemic resulted in a sudden and unprecedented increase in e-commerce transactions, presenting an opportunity for AIenabled anti-fraud and cybersecurity companies. Private vendor Sift, for instance, reported assessing $250B worth of transactions for risk — double 2019’s total. Source: Sift 6 AI companies analyze numerous data points, including location data, device ID, social data, and others in real time and reduce false positives while flagging suspicious transactions. The space saw 2 new unicorns recently: • Sift raised $50M in Q2’21 at a $1B valuation and acquired early-stage startup Chargeback to expand its product e-commerce portfolio to include dispute management. • Forter raised $300M at a $3B valuation from Bessemer Venture Partners, Sequoia Capital, Salesforce Ventures, and others this year. Apart from the pandemic-induced surge in online transactions, online merchants were also met with the European Commission’s deadline for new customer authentication standards in 2020. To comply with the second Payment Services Directive (PSD2) Strong Customer Authentication (SCA), vendors now have to authenticate at least 2 of of the following 3 factors: • Something a customer knows (like a password) • Something a customer has (device ID, etc.) • Something a customer is (voice, fingerprinting, and other biometric authentication) Companies like BioCatch offer validation for the third criteria through behavioral biometrics authentication instead of voice or fingerprinting. Biocatch analyzes online behavioral data such as unique data entry patterns, hand tremors, left/right-handedness, scrolling patterns, and more. 7 Source: BioCatch In our Cybersecurity in Retail Tech Market Map report, we analyze the various vendors and tools available to help retailers to protect critical operations and customer data, as well as how brands are navigating the fragmented and quickly evolving cybersecurity market. 8 Retailers will invest more in microfulfillment infrastructure PepsiCo, Walmart, and others have piloted or scaled their microfulfillment operations in response to Covid safety protocols and an uptick in e-commerce orders. Before Covid-19, supermarkets and retailers were already facing pressure from tech giants like Amazon to fulfill orders within hours. The pandemic accelerated the trend, as e-commerce purchases skyrocketed. Robotic micro-fulfillment centers promise to make e-commerce profitable for supermarkets while helping them engage directly with customers. Micro-fulfillment centers (MFCs) are mini, vertically stacked warehouses that can fit within an existing retail space. The entire “mini warehouse” is less than 10K square feet (in some cases around 3,000 sq. ft), while traditional warehouses can be the size of a football field. Shelves are vertically stacked to save space, and can be installed inside existing supermarkets, building basements, or even parking garages. Ground robots move between aisles to fetch the items in an order and hand them to a human worker for final packaging. AI-backed software is used to decide where goods are placed on shelves, prioritize tasks, and send navigation instructions to ground robots. 9 Novastore: Alert Innovation’s concept of a self-service supermarket MFCs have the most potential in grocery retail, although they are gaining traction across other consumables categories like health & beauty. This year, Walmart announced that it is scaling its micro-fulfillment projects with different technology vendors including Alert Innovation, Dematic, and Fabric (previously CommonSense Robotics). “They [micro fulfillment centers] move a significant amount of the picking off the sales floor, allowing us to do more within the box. And one of these fulfillment centers can serve a large area spanning multiple communities. And we’re now moving to scale these locations and we expect to have over 100 of these within the next couple of years. And at some stores, we’ll carve out existing space for them. At others, we’ll add-on.” — WALMART CEO DOUG MCMILLON, Q4’21 EARNINGS CALL 10 Dematic also partnered with PepsiCo last year to keep up with online orders. PepsiCo announced that the “fully automated fulfillment approach improves Covid safety, reduces the costs of floorspace, and expedites the picking process.” Another vendor, Takeoff Technologies, which works with retailers like Albertsons, Ahold Delhaize, and Carrefour, announced plans to have 40 automated MFCs up and running by the end of the year. The cost of setting up and running these centers in partnership with startups is undisclosed. But an operating assumption is that, since these fulfillment centers will be in high-density urban pockets, last-mile delivery would cost less — or nothing at all, if consumers choose to order online and pick up in-store. 11 Supermarkets will tackle food waste reduction with AI Supermarkets are partnering with AI vendors for markdown optimization, fresh food inventory management, and meeting corporate sustainability goals. Food waste accounts for billions of dollars in lost revenue opportunity for grocery retailers. For 31 fresh vegetable categories alone, supermarket waste amounted to 6.2B pounds in 2016, according to the US Department of Agriculture. The pricing optimization challenges for fresh produce are unique: product shelf life can be as little as a day or two; discolored produce has a smaller likelihood of getting picked; and consumers are statistically more inclined to buy products with a longer expiration date. A markdown optimization strategy incentivizes customers to purchase less fresh produce for the benefit of lower costs. Wasteless uses an AI technique called reinforcement learning to predict if an item will sell, and adjusts prices based on 40+ variables, including sale strength of a brand, demographic area in which the item is sold, day and hour of the sale, competing products, and demand curve, among others, per a 2017 patent application. 12 The vendor offers retailers electronic shelf tags along with its pricing engine. Wasteless’ tech is in trial with the METRO group in Poland. Other companies, like Farmstead and Afresh, use historical sales data to predict demand. Grocery supply chain startup Afresh raised $25M in 2020 from Baseline Ventures, Innovation Endeavors, Impact Engine, Maersk Growth, and Food Retail Ventures. “We’re looking at sales and shipments, but also using an algorithm to assess what we think is in the store, with a confidence interval that includes how perishable the item is.” — AFRESH CEO MATT SCHWARTZ, IN AN INTERVIEW WITH FORBES Using AI, demand models can quickly adapt to unexpected events like the pandemic. For example, Farmstead uses AI to optimize its “farm to fridge” digital operations. The company spun off its FreshAI tech as a B2B SaaS solution for other retailers after a successful internal pilot that cut food waste to under 10%, compared to the average 30%-40% at supermarkets. Apart from improving margins, retailers have vested interest in using AI to comply with UN sustainability goals, and improving their investor ESG (environmental, social, and governance) scores. Companies are committing to more sustainable practices, with discussions around ESG reaching an all-time high during the pandemic. 13 Kroger has partnered with Project Delta, a spin off from Alphabet’s moonshot lab, to meet its “Zero Hunger, Zero Waste” goals with AI. Delta works with Kroger and participating food banks to make the processing of routing excess food for donation more efficient. 14 Retailers will double down on first-party data strategy in a post-cookie world Third-party cookies are dying. AI-powered consumer data platforms, which combine first datasets to create a unified shopper profile, are stepping in to fill the gap. In January 2020, Google announced that it would ban thirdparty cookies on the Chrome browser by 2022. Safari and Firefox announced similar plans earlier in lieu of increasing consumer privacy concerns and government scrutiny. This changes consumer targeting and retargeting strategies for all companies, including retailers and advertising agencies that work with retail clients. Google is building alternative targeting strategies for advertisers and publishers — such as its Privacy Sandbox, which uses federated learning to group consumers into cohorts that can be targeted without having to identify individuals. But the demise of cookies has renewed interest in first-party data. 15 “Every conversation, I can say, I’m a part of, it includes first-party data now. And the need, quite frankly, for brands to own this relationship, not platforms, to own the data and the relationship with the consumer.” — WENDY CLARK, DENTSU CEO, Q1’21 EARNINGS CALL First-party data is anything a retailer collects directly from the consumer, including app usage data, online browsing behavior, existing CRM data, and point-of-sale data. Customer data platforms (CDP) — or vendors that offer solutions to ingest all of these different datasets to create a unified shopper profile — are gaining traction, with the space seeing some notable acquisitions in 2020. For example, Bloomreach — which is valued at $900M and works with brands including Albertsons, Staples, Bosch, Puma, and FC Bayern Münche — acquired CDP Exponea. 16 Twilio, which builds communication APIs for industries including retail (with clients like LensDirect, Shopify, and eBay) acquired CDP unicorn Segment for $3.2B. Image source: Segment.com AI-powered CDPs help retailers unify and deduplicate shopper profiles, cluster similar shoppers together, and generate advanced business and operational insights. For example, ActionIQ stitches together first- and third-party data to personalize customer experience. The company is backed by Smart Money VCs Sequoia Capital, FirstMark Capital, and Andreessen Horowitz. It raised a $67.6M follow-on Series C round in Q1’21. ActionIQ’s clients include Shopify, Neiman Marcus, and Michael Kors. With more regulations around consumer data privacy and the sunsetting of third-party cookies, expect more retailers to be deliberate about the data they can directly collect to power recommendations. 17 Auto tagging will become a must-have AI tool for online retailers Companies like Thredup, Everlane, The Yes, and others are building the most extensive product taxonomies in fashion retail with AI. Product taxonomies are hierarchical classifications (for instance, Women’s » Apparel » Long gown) assigned to items to make them easily searchable by consumers. For consumers, a well-designed taxonomy improves the product discovery and browsing experience. This is the top motivator for retailers — a robust taxonomy improves sales. With retailers carrying more items and consumer browsing behaviors changing, manually maintaining a dynamic taxonomy is unscalable and expensive. Artificial intelligence is now changing the way retailers maintain their catalogs. Source: Syte.ai 18 Natural language processing and computer vision automatically generate tags based on the product description and visual attributes. Lily AI, an early-stage vendor in the space, claims its algorithms were trained on a proprietary data set of 1B data points that were manually tagged by retail experts. It has a product dataset of 15,000 tags that it can attribute to items. Lily differentiates itself by offering retailers psychographic profiles of consumers based on the tags and taxonomies of products they browse. “Do they love boho chic? Do they have a short torso or long arms? Do they live for sequins? Learning this starts with better, more accurate fashion tagging,” the company explains on its blog. Lily currently works with Thredup, Bloomingdale’s, Everlane, and others. It raised $12.5M in 2020 from New Enterprise Associates and others. Meanwhile, Israel-based Syte uses computer vision to develop detailed tags for products, which are then used to create more complete product descriptions on e-commerce sites. The company raised $30M last year from NAVER and others. Taxonomies can be particularly challenging for secondhand retailers, where the items sold in are not consistent or uniform. Secondhand retailer White Rose used Syte’s visual AI to transition from brick-and-mortar to e-commerce during the pandemic, and claims shoppers converted at a rate 6.65X higher than the average. Meanwhile, other companies in the space like Glisten (below) offer APIs where retailers can upload their data for thousands of products and receive detailed tags in a couple of hours. 19 Source: Glisten AI AI-powered e-commerce platform The Yes raised its first funding round in 2018 and created buzz for its star-studded founding and engineering teams. The company officially launched its consumer shopping app in 2020 with 140+ brands including Gucci, Prada, Vince, and others. CEO Julie Bornstein said The Yes’ extensive work on taxonomies was inspired by similar efforts in the music industry by Spotify and Pandora in classifying songs. 20 2020 also saw big tech companies like Facebook advance the state of image recognition and product identification. Facebook’s AI team launched GrokNet, an image recognition system to identify attributes of products listed on its marketplace. “Product tags make it easier for people to find the items featured in content in their feed. So, in the future, let’s say you’re eyeing a pair of sneakers, but only the brand is tagged. Normally, you might have to search through the brand’s website to find the product. Our AI system would automatically scan through the product catalog and identify the exact pair of shoes to enable a more convenient shopping experience.” — FACEBOOK Apart from increased retention and sales, a standardized and granular taxonomy improves competitive intelligence for retailers, who can compare their assortment with competitors that may use a different classification to describe the same product. 21 AI will power hyper-local inventory at the store level Retailers are using AI to group similar stores together and execute localized inventory management strategies based on demographic trends and real-time market conditions. Retailers use inventory management and optimization tools to determine what products to stock up on, in what quantities, and at which location. An efficient plan minimizes markdowns and stockouts while maximizing revenue. Assortment planning has typically relied on historical transaction data. However, during the pandemic, consumer behaviors changed drastically in a short period of time, rendering historic data useless. This renewed retailers’ interest in AI-driven assortment planning tools that can factor in market conditions, real-time purchase behaviors, climate events, local demographic information, store size, and assortment of competing retailers. “The pandemic accelerated the customer focus to hyperlocal. When we look at Google Search trends from last year, consumer searches for specific items in stock increased 8,000% in the US; searches just for in-stock items went up 700%, and curbside went up 3,000%.” — CARRIE THARP, VP OF RETAIL AND CONSUMER, GOOGLE CLOUD 22 Louis Vuitton called the last 18 months “transformational” for retail in underscoring the importance of leveraging data. The fashion brand inked a deal with Google to automate supply chain and demand forecasting, personalization, and more. One of the use cases of AI in assortment planning is “clustering” similar stores together, with each cluster having its own dynamic assortment plan. This is particularly useful for larger brands such as Levi’s that want to execute investor strategies at the scale of hundreds of stores, while still catering to hyper-local consumer needs. “At our stores, AI is enabling local stores in China and the U.S. to better curate their assortments by predicting demand based on the specific profile and preferences of consumers in the vicinity of each store, which will thereby optimize the profitability of these smaller mainline doors.” — CHARLES BERGH, LEVI STRAUSS & CO CEO, Q3’20 EARNINGS CALL Levi’s partnered with Lynx Analytics to cluster 300+ stores across China into groups based on consumer purchasing behavior and product preferences. 23 Source: Lynx Analytics AI software can also simulate the impact of adding or dropping SKUs to categories, not only based on data from a single store, but from product trends and consumer behaviors observed across a network of stores that use the same software. Early-stage company Hivery claims it can reduce the category management process — including predicting demand for a specific category of products, planning their placement on store shelves, and optimizing promotional strategies — from months to minutes. Hivery’s tech was developed in the digital lab of the Australian government’s science agency CSIRO. Its assortment planning and category management tools are used by Walmart, Coca-Cola, and others. 24 Checkout-free solutions will become more accessible for retailers Whether retrofitting the entire store with checkout-free tech or deploying smart carts, big tech companies and smaller private vendors are broadening access to cashierless tech for retailers. Amazon triggered a checkout-free store frenzy when it announced plans for the Amazon Go store in 2015, allowing shoppers to grab items and walk out. The company had no public plans to sell its tech as a service to other retailers, and had been tight-lipped about the operations, success, and pain points — only revealing that it uses sensors, cameras, computer vision, and deep learning algorithms. That changed in 2020, when Amazon launched “Just Walk Out,” a business unit focused on selling cashierless tech to other retailers. Source: justwalkout.com In the interim, the market also saw several private vendors step in to sell the tech to retailers, including Grabango, Standard Cognition, Aifi, and others. 25 Although the tech is still in its infancy and the cost-benefit details for retailers are opaque (some analysts estimate the costs of tech deployment in each Go store to be around $1M), Amazon deciding to enter the market as a technology partner gives it steam. Smart carts are also gaining traction as a bolt-on solution. For example, New Zealand-based smart cart maker Imagr is piloting its Halo Cart — which has built-in computer vision to recognize items — in Japan’s Oasis supermarket chain. Moreover, Kroger announced in January 2021 that it is using Caper’s computer visionpowered smart shopping cart to pilot its KroGo cart. Amazon also launched its own smart cart called Dash Cart in 2020. Dash Carts are available in Amazon Fresh grocery stores and in select Whole Foods locations. Cashierless checkout could increase convenience and speed for shoppers, enable data gathering on consumers and their shopping habits, and eliminate shopper theft. The systems can also reduce labor costs by eliminating cashiers. However, the retailers and vendors implementing the tech are careful to note that those employees can be better deployed for more sophisticated tasks elsewhere in the store. 26 Additional reading This report was created with data from CB Insights’ emerging technology insights platform, which offers clarity into emerging tech and new business strategies through tools like: • Earnings Transcripts Search Engine & Analytics to get an information edge on competitors’ and incumbents’ strategies • Patent Analytics to see where innovation is happening next • Company Mosaic Scores to evaluate startup health, based on our National Science Foundation-backed algorithm • Business Relationships to quickly see a company’s competitors, partners, and more • Market Sizing Tools to visualize market growth and spot the next big opportunity If you aren’t already a client, sign up for a free trial to learn more about our platform. 27
关注微信
