Artificial Intelligence in Retail, Part 2: What Retailers Are Doing with AI

Introduction

AI continues to evolve and pervade business functions and applications across industries. In retail, AI’s use has moved well beyond warehouse management and e-commerce algorithms. Companies are now deploying it to help solve far more complex problems and facilitate functions across their businesses.

In this report, our second in the Artificial Intelligence in Retail series, we outline some of the key areas in which 30 global retailers are using AI across business functions. We also examine in detail how two of the largest global technology-led retailers, Amazon and Alibaba Group, are using AI in alignment with our proprietary CORE framework. Introduced in our first report in this series, which can be found here, the CORE framework illustrates the value of AI for retailers in four areas: communication, optimization of pricing, rationalization of inventory and experiential retail. Finally, we take a deeper look at how some other selected major retailers are developing and deploying AI for both business functions and customer-facing applications.

AI Use Centers on Supply Chain, In-Store and E-Commerce Operations

Most retailers tend to use AI for supply chain functions such as inventory management, demand forecasting, automatic replenishment, automated warehouse operations and fulfillment. Many are also now using AI for in-store and e-commerce functions such as personalization, product and sizing recommendations, price optimization, and promotions.

Out of the 30 retailers we studied, 70% or more are already using AI for supply chain and customer-facing functions. While Amazon, Alibaba and JD.com are the only retailers we studied that have developed proprietary smart speakers, several other retailers have partnered with Google or Amazon to allow customers to shop via voice command through the companies’ respective smart speakers.

We examine some of these companies’ AI initiatives in more detail below.

Amazon

Amazon continues to lead technology innovation in retail, primarily because it positions itself as a technology company rather than as a retailer. By using AI across its internal business as well as in customer-facing functions, Amazon covers all four quadrants of our CORE framework for AI in retail (shown below). Amazon’s Alexa—a personal assistant activated and run by voice commands—is the firm’s best-known example of AI in action.

Some of Amazon’s other popular inventions are:

Delivery drones: These aerial vehicles deliver packages to customers in 30 minutes or less from the time they an order is placed.

Warehouse robots: The company’s robot fulfillment systems and network were created by its subsidiary, Amazon Robotics, which was known as Kiva Systems before Amazon acquired it in 2012. Kiva catered to several other retailers, including Walgreens, Gap and Staples, until their contracts expired. Amazon did not renew the contracts, giving it complete control over Kiva’s inventions.

Amazon Go: Amazon’s unstaffed, cashierless store format allows shoppers to pick up items they wish to purchase and walk right out of the store. Customers scan their phones when they enter an Amazon Go store and are automatically charged for the items they have picked up when they leave. Thousands of cameras and sensors around the store track which products customers remove from shelves and place in their basket or bag.

Amazon also deploys AI in some functions that receive less attention than those noted above. For example, the company uses AI in its e-commerce site operations and for inventory planning and management. These two areas may seem like obvious fits for AI usage, but they are generally talked about less when industry watchers and analysts discuss Amazon’s use of AI.

Inventory management and retail operations: Amazon is increasingly using AI to make better decisions regarding retail operations, according to a recent report by news agency Bloomberg. It applies the technology to help forecast demand, order stock and negotiate prices with suppliers, for instance.

Amazon runs its retail operations alongside its marketplace, which features numerous third-party sellers. According to Bloomberg, the company’s retail teams were prompted to adopt AI because the marketplace produced twice the operating profit as the retail business (a source quoted by Bloomberg said that the retail division did not make a profit in several international markets). Technology has enabled Amazon to consolidate its retail and marketplace operations as well as automate many tasks that were previously performed by human staff.

Product pages, deals and promotions: Globally, Amazon is using AI to create a seamless user experience on its website and app, with advertisements and suggestions aiding product search and discovery.

In India, Amazon is using AI and machine learning to help sellers categorize their products appropriately, track which products sell best during holidays and festivals, and formulate promotions that will prompt higher sales of those products. The retailer is also working on using AI to improve image quality and product descriptions, identify and remove fake reviews, translate millions of product pages, and provide content in India’s regional languages.

Amazon has also been actively investing in AI startups. It launched the Alexa Fund in 2015 to focus on early-stage funding of startups that develop machine-learning and AI solutions. Below is a list of startups that Amazon has acquired or invested in since 2009.

Amazon is not the only major retail player in AI: Chinese marketplace platform and retailer Alibaba has been working on several AI initiatives that put it on par with, or even ahead of, Amazon in terms of AI innovation.

Alibaba

Alibaba, China’s very own Amazon and the purveyor of “New Retail,” has embraced AI. Alibaba Chairman Jack Ma first introduced the company’s New Retail strategy in a letter to shareholders in October 2016, stating that “pure e-commerce will be reduced to a traditional business and replaced by the concept of New Retail―the integration of online, offline, logistics and data across a single value chain.” Alibaba is now using AI across several functions to implement its New Retail strategy.

Tmall Genie: Alibaba’s Tmall Genie smart speaker is similar to Amazon’s Echo, which works with the Alexa AI assistant. Alibaba’s AI speaker platform, called AliGenie, runs on voice commands, and Alibaba intends to expand its utility beyond retail into the education and travel sectors as well. Tmall Genie works as a remote control for smart home products: it switches on devices, reads children bedtime stories and orders food for delivery. Alibaba is also encouraging supermarkets to adopt the speakers to answer shoppers’ in-store questions, such as “Where can I find the sugar?”

tion via an app that works with the smart speaker. Users need to download the app, which links with the speaker, and then place their phone on top of the speaker using a fixture from the company.

The app has many uses. For example, users can scan a book cover with the app in order to have it read the book to them. They can also have it read medicine labels out loud, which is particularly useful for those with visual impairments.

Dian Xiaomi: Alibaba’s AI-driven customer service chatbot is called Dian Xiaomi. It is especially helpful during the retailer’s annual Singles’ Day shopping festival, when customer queries increase enormously versus a regular sales day. According to Alibaba, the chatbot can address more than 3.5 million customer queries in a day and is capable of understanding more than 90% of queries. The chatbot can also analyze customers’ emotions and flag orders that are of higher priority.

E-Commerce Brain: Alibaba’s E-commerce Brain is AI software that studies customer behavior and personalizes recommendations, deals and offerings based on its analysis. Alibaba claims that its software helped generate 6.7 billion personalized shopping pages for its Taobao and Tmall e-commerce sites during its 2016 Singles’ Day shopping festival, and that those pages led to a 20% higher conversion rate compared with nonpersonalized pages.

ASSC: Alibaba’s AI-powered supply chain technology overlaps the “optimization of pricing” and “rationalization of inventory” quadrants in our CORE framework. ASSC uses predictive analytics to forecast demand, replenish and allocate inventory, decide product assortment and devise pricing strategies. The system uses both historical and real-time data and can be applied to both newly introduced and seasoned product offerings.

Cainiao: Alibaba’s intelligent logistics and warehouse management network, Cainiao, uses AI and data analytics for parcel sorting and delivery. The firm claims that Cainiao enabled 70% of parcel deliveries in China as of August 2016, processing an average of 42 million deliveries per day. Alibaba says that Cainiao’s algorithms enabled the company to reduce the use of packing materials by more than 10% by analyzing product and packing material sizes. The network’s route optimization models have enabled the company to cut vehicle use by 10% and travel distances by 30%.

FashionAI: FashionAI is a digital consultant that uses AI to source fashion and dressing tips data from stylists on Alibaba’s Taobao and Tmall platforms. During the 2017 Singles’ Day shopping festival, Alibaba provided smart dressing rooms at 13 stores, where shoppers could try on clothing with embedded RFID sensors. The dressing rooms were equipped with interactive screens that identified the clothing through the sensors and then provided recommendations based on the shoppers’ choices. Shoppers could choose the items they liked and push a button to notify store assistants to bring the selected clothing to their dressing room.

New Retail stores: Alibaba’s Hema supermarkets exemplify the firm’s push to integrate the online and offline channels through New Retail stores. Customers use the retailer’s mobile app to research products in the stores, and then pay through Alipay. The stores offer in-store fulfillment of online orders, interactive kiosks for browsing the online selection in-store and purchasing products, free delivery of orders, and a dining area where shoppers can have in-store chefs cook the foods and fresh produce they have selected.

Alibaba recently partnered with Sun Art Retail Group’s RT-Mart to replicate the Hema store model at RT-Mart hypermarkets. Through the partnership, Alibaba will provide technology to help improve store layouts, accelerate digitalization and increase efficiencies in inventory management at RT-Mart stores.

Alibaba’s AI initiatives are not limited to retail. They cover a wide spectrum of applications, including:

City Brain: Alibaba’s City Brain is a smart city platform that enables the digital transformation of cities’ public services, such as transportation, safety, governance, security and energy. City Brain’s systems process vehicle and city data to optimize traffic management, detect incidents and notify emergency services when needed. The systems also track residents’ behavior to devise ideas for city planning and development.

Tmall Genie integration with smart car systems: Alibaba is working on integrating its smart speaker with smart car systems from BMW, Volvo, Audi and Daimler in China. Tmall Genie will be able to connect to the cars and check fuel levels and whether the cars are locked. It will also be able to check travel routes, recommend departure times, and remotely activate heating or air conditioning.

AI chip development: Neural-processing units (NPUs) are computer chips built specifically for machine learning and AI algorithms and tasks. Global IT vendors such as Intel and Nvidia already make such chips, but Alibaba is developing its own NPU chipset, called Ali-NPU, to be used in its business operations that involve AI and in its devices that use AI to talk to each other. Alibaba has acquired chip manufacturer Hangzhou C-SKY Microsystems to accelerate the development of Ali-NPUs.

Alibaba’s biggest investments have been primarily in e-commerce firms, but in April 2018, the company poured $600 million into a facial recognition startup called SenseTime. The startup will partner with Alibaba on its smart city initiatives and provide applications for intelligent surveillance, transportation and urban management.

In October 2017, Alibaba launched a program called the DAMO (which stands for “discovery, adventure, momentum and outlook”) Academy to focus on R&D of new technologies such as AI, the Internet of Things and smart city projects. Chairman Jack Ma stated that Alibaba will dedicate $15 billion to the DAMO Academy, which will open seven technology labs around the world.

Who Else Is Doing What?

Several other retailers are using AI across business functions and for customer-facing touch points. We outline some of the most notable uses below.

Carrefour

• Google Shopping: Carrefour has partnered with Google to sell on the technology firm’s shopping platform as well as through its voice-based Assistant. The partnership also involves Carrefour engineers working with Google Cloud AI experts to develop new customer experiences.

eBay

• Pricing and inventory: eBay uses pricing and inventory algorithms to recommend prices based on local demand and to notify sellers when stock levels are low.
• Listing optimization: AI is helping eBay sellers list their products appropriately and optimize images to gain more attention.
• ShopBot: eBay’s ShopBot chatbot helps shoppers browse and make purchases and delivers recommendations through social platforms and messaging applications.
• Image Search and Find It on eBay: This application uses AI and machine learning to allow shoppers to search for products using images on their smartphones. By uploading an image into eBay’s search bar, shoppers can find products that are similar to the one pictured.

JD.com

• Unstaffed stores: JD.com has opened several unstaffed stores where shoppers can use their smartphones to shop for products. The stores are similar to Amazon Go stores and Walmart’s Scan & Go offering.
• Unstaffed warehouse: Robots and robotic arms receive, move, store, sort, pack, categorize and dispatch products in JD.com’s only unstaffed warehouse. A handful of staff are present, but the operations are carried out by AI-powered robots to a large extent.
• Smart carts: The carts in JD.com’s 7Fresh fresh-food supermarkets guide shoppers to the aisles where the products they are looking for are located.
• Collaboration with Fung Retailing: JD.com is collaborating with Fung Retailing to build an AI-driven retail center. This initiative is aimed at converting JD.com’s retail business into an open platform to support other retailers by providing “retail as a service.”

Ocado

• Smart Platform: Ocado’s Smart Platform is an end-to-end online grocery system that includes an automated warehouse and accompanying software applications. It was first deployed in Ocado’s Andover, England, warehouse, which spans 240,000 square feet over five stories and has an operating capacity of 70,000 orders per week. Groceries and other products are stacked in hivelike shelves that run many levels high. Overhead robots that hang from roof beams pick and move products across the facility. The warehouse, which uses machine learning to operate functions, stores thousands of kinds of products across several temperature systems. The technology is used to move products across the floor or give them to a human picker who is fulfilling orders.
• SecondHands Cobot: Ocado is developing a collaborative humanoid robot that uses AI, machine learning and advanced vision to assist human workers with various tasks.
• Contact center applications: Ocado used information from Google’s TensorFlow—a library for machine intelligence—to develop capabilities for its contact center. The application is able to tag and categorize customers’ emails according to priority. It can sort through emails, identify those that are feedback, and identify whether a sender is happy or has asked a question that needs to be addressed.

Sephora

• Color IQ: Sephora’s in-store machine learning–driven Color IQ device uses a sophisticated camera to scan skin color. The device analyzes the data and generates a unique Color IQ number to help the user find the right lipstick, foundation and concealer shades.

Tesco

• Inventory management: Tesco uses AI to forecast and replenish inventory by store, taking into account a number of variables, such as weather and regional demand. It also uses computer vision algorithms through its in-store static cameras to replenish store shelves.
• In-store routing algorithms: In-store routing algorithms help shorten the distances that Tesco Online personal shoppers must walk when fulfilling orders.

Walmart

• Store No. 8 incubator: Walmart launched Store No. 8 to “incubate, invest in and work with other startups, venture capitalists and academics to develop its own proprietary robotics, virtual- and augmented-reality, machine-learning and artificial intelligence technology.”
• Walmart also uses machine-learning algorithms for organizing inventory data, pricing items, optimizing routes for delivery and solving other operational problems.
• The retailer is working on a facial recognition algorithm to recognize customers’ facial expressions and identify their levels of satisfaction.

What We Think

AI has become ubiquitous, and many retailers have adopted the technology to a limited extent. Many others, however, are already using it widely across business functions, exploring a broader scope of possibilities and perhaps giving themselves a stronger competitive edge.

AI technologies have proven their efficiency and it is clear that retailers need to adopt AI, whether they choose to develop a majority of their AI technology in-house, like Alibaba does; mix in-house development with acquisitions of startups that have established AI competencies, like Amazon does; or simply rely on specialist AI technology vendors for specific functions.

Retailers can look to our CORE framework to understand which specific areas of their businesses they can focus their AI efforts on in order to engage with customers more effectively.