Discovery AI is born, here is Google’s artificial intelligence at the service of retail

Listen to the audio version of the article

Four new algorithm-based technologies specifically designed to help retailers transform (and make more efficient) in-store shelf inspection processes and improve e-commerce sites by delivering customer experiences more fluid, natural and personalized online shopping. The announcement signed by Google Cloud on Friday 13 January may not go down in history or even mark the future of the Californian company, but there is no doubt that the “Discovery AI” solutions being presented in New York can be a further turning point for the retail industry. Carrie Tharp, VP of Retail and Consumer of the BigG division, is absolutely convinced of this, according to which “the changes of recent years have reshaped the retail landscape and the tools retailers need to be more efficient, more attractive for their customers and less exposed to future shocks. Tomorrow’s leaders – she specified in a note – will be those who face today’s most pressing challenges, in-store and online, using the most advanced technological tools, such as machine learning ”.

Improve product availability

Finding yourself with limited or even no inventory on the shelf is one of the most heartfelt concerns of retailers: an analysis by NielsenIQ tells us in this regard that in 2021 alone, the lack of product in stores cost US retail operators something like 82 billion dollars of lost sales. For years, new technological solutions have been experimented and tested to detect and differentiate items more rationally, but more often than not the effectiveness of these solutions has often been limited by the scarce resources available to create reliable AI models. Google Cloud’s artificial intelligence (AI Vision) addresses this need with better visibility into what shelves actually look like to help you locate where supplies are needed faster. Based on the Vertex engine that scans billions of unique entities present in the database of the Mountain View company, the solution in question is powered by two machine learning systems (one for product recognition and one for tag recognition) and identifies the products based exclusively on the visual and textual characteristics of a product (comparing images taken from different angles and points of view) translating this data into immediately usable insights.

AI transforms the digital shopping experience

The old adage of e-commerce sites involved sorting the results of products based on the best sellers in each product category or on rules pre-established by people. Making the online browsing and catalog item evaluation experience faster, more intuitive and more satisfying for buyers is the dream of every retailer and the proposal that Google Cloud is putting in place today is a new navigation function powered by artificial intelligence (localized in 72 languages) which leverages the historical data analysis capabilities of machine learning to set the optimal order of products for each page of an e-commerce site once users choose a category, optimize how and which items are displayed for accuracy, relevance and likelihood of sale.

More personalized searches thanks to machine learning

Research commissioned by Google Cloud found that 75% of consumers prefer brands that purposefully manage customer interactions, and 86% consider brands that understand their interests and preferences. From these indicators, the BigG engineers started to develop a function (based on artificial intelligence) capable of processing the results that a user obtains when browsing a retailer’s website in a personalized way. More specifically, the AI ​​behind this feature is a “recognizer” of product models and determines the consumer’s tastes and preferences by examining their purchasing behavior on an e-commerce site (they are not then map the activities conducted by the user as an account holder on Google): when a product matches the preferences already categorized, the algorithm moves it to the relevant positions in the search and navigation rankings to obtain a personalized match.

More targeted recommendations for consumers

What information should you post and highlight on your websites? How to organize them effectively and how to coordinate relevant and “tailor made” content? Questions that many retailers probably continue to ask themselves and which Google Cloud tries to answer with the Recommendations AI solution, subject to updates that add new tools for customizing e-commerce sites. Thanks to machine learning, in fact, the page-level optimization function now offers “merchants” the possibility to dynamically decide which product suggestions to show uniquely to a buyer, improving their engagement and conversion rates. Another model, created in collaboration with DeepMind, instead exploits the self-learning capabilities of the algorithms and combines the product categories of an e-commerce, the prices of the items, the clicks made and other information related to the digital shopping experience to find the right balance between long-term satisfaction for users and increased revenue for resellers.