The Role of Machine Learning
The role of machine learning in Alibaba’s product development process is immense. Part of the reason Alibaba’s machine learning capabilities are so powerful is the company’s ability to integrate data throughout its ecosystem, incorporating transaction data from shopping platforms as well as credit data from Alipay, and video consumption data from Youku, amongst many other datasets. This provides Alibaba with a broad set of behavioral information to power recommendations, and means the company can suggest both online and offline local products and services in real-time.
Alibaba is using machine learning for many aspects of its business in order to better serve its retail partners and customers. These include tools to personalise recommendations, handle customer enquiries, target advertising, and manage inventory. Alibaba’s B2C retail platform Tmall offers “Smart Selection”, an algorithm that helps recommend products to shoppers and communicates this to retailers to increase inventory and keep up with demand.
During the 2016 Singles Day, Alibaba achieved personalization for all of its retail marketplaces for the first time, from the Taobao and Tmall homepages, to other promotion pages and product detail pages. The result of this was a conversion rate 20% higher than that of non-personalized pages.
In an environment of fierce competition with retailer JD.com, Alibaba is seeking to innovate, make strategic acquisitions and expand its technological capabilities. Alibaba’s focus on innovation was underscored in 2017, when it announced a $15bn investment into a new research institute. The DAMO institute, dedicated to technological research, has a strong focus on Machine Intelligence, with the aim of developing innovations not only for Alibaba’s core e-commerce business, but for many other applications.
Alibaba’s “New Retail Strategy”
Alibaba’s “New Retail” strategy, exemplifies the company’s desire to cater to future consumer behavior and draw customers deeper into its ecosystem. The aim is to integrate online and offline shopping experiences, where interactions among consumers, inventory location and retail space are enhanced by leveraging mobile and big data. The goal is to reinvent retailers by increasing their catchment area online and offline and digitizing operations.
This initiative can be seen across both Alibaba-owned entities (including grocery store Hema and department store Intime), as well as third-party retailers.
For example, Alibaba’s Hema stores are used for regular shopping and in-store dining but also as warehouses for online orders. Alibaba uses online and offline transaction data to personalise recommendations, and geographic data to plan efficient delivery routes.
Similarly, Intime has benefited from Alibaba’s infrastructure to innovate in areas from membership to payments to logistics. In some store locations, Alibaba experimented with its automated fashion consultant FashionAI. A screen would scan item tags on products that customers were holding, then use machine learning to make suggestions on what to pair the item with.
Alibaba is also turning independent convenience stores digital through its platform “Ling Shou Tong” by equipping stores with technological tools. In return for providing analytics and a digital inventory system, Alibaba receives data on consumer behavior and the ability to use these stores as fulfilment centers.
By building out its omni-channel capabilities, Alibaba is becoming more ingrained in the daily lives of shoppers and retailers, gathering increasingly rich data which will help solidify its competitive positioning.
The next step
Alibaba may be able to do more with its machine learning capabilities and extensive datasets. Looking to Amazon, there may be an opportunity to capture an increasing share of consumers’ wallets by selectively verticalizing through launching private label products.
Given Alibaba’s data on consumer preferences, its ability to target and personalize, and its inventory management systems, the company has a unique place from which to launch new products. Alibaba understands better than most which products will be popular amongst shoppers and how to service them efficiently, optimising advertising and pricing.
Investing in private-label apparel may also give Alibaba the chance to experiment with new experiences and enhanced personalization. For example, Alibaba could trial “try before you buy” as Amazon did with its Prime Wardrobe offering whereby customers receive a package of clothing and only pay for what they keep. I also see potential in enhanced styling to deepen consumer attachment. Looking to StitchFix, Alibaba could also use its FashionAI to scale personalised styling, potentially with some human interaction. This may be a powerful proposition in the context of JD.com’s other efforts to increase personalisation such as its “white glove” last mile delivery service.
Nevertheless, there are clear downsides to these suggestions, including capital intensity and complexity of managing manufacturing processes and inventory. Moreover, should Alibaba pursue private-label brands more aggressively and cannibalise other retailers, there is a risk that they will erode the trust of their B2C clients who may seek to distribute elsewhere.