AI determines which online stores are visible to you, overlooking many of them.

      TL;DR: A study conducted by AI commerce company Recomaze analyzed 9,720 ecommerce stores, running 58,320 product queries through Google Gemini. In 60% of cases, no store was recommended, and when there were recommendations, they spanned over 50,000 brands.

      Shoppers are increasingly turning to AI assistants for purchasing advice rather than just for basic inquiries. This recent study indicates that the majority of online retailers are not included in the responses provided by these assistants. Recomaze’s research involved executing six purchase-intent queries for each of the 9,720 ecommerce stores, totaling 58,320 evaluations. In 60% of instances, no store was recommended at all. When a store was mentioned, it only occurred 14% of the time, while competitors were highlighted or no store was named in the remaining cases.

      As Amazon, Google, and others enhance the role of AI in shopping, the question of "what should I buy" may result in a limited list of store names rather than extensive links to options. Being included in such a selective list equates to a new form of visibility akin to ranking on the first page of search results. The study posits that many product catalogs are designed for human consumers and search engines like Google, rather than optimized for AI that must interpret and recommend products for shoppers.

      An interesting aspect of the study is the diversity of the brands recommended. Recommendations were not dominated by a few large companies; instead, they were distributed among 50,287 unique brands. The top ten recommended brands accounted for roughly 4% of all mentions, with the leading brand, Etsy, appearing in just 1.3% of requests. In one example, a request for cocktail capsules highlighted Bartesian, a niche brand, over larger retailers like Walmart and Target, indicating that AI product discovery favors a long-tail approach rather than automatically favoring larger stores.

      The study found that category type was the strongest indicator of store visibility. Categories driven by visual appeal, such as home and living and apparel, had the highest rates of invisibility at approximately 74% and 67%, respectively. Conversely, food and beverage had a visibility rate of around 52%. This discrepancy arises because a text-based engine finds it challenging to assess the aesthetics that influence buying decisions for clothing and home goods, leading it to revert to larger, generic retailers, whereas products with easily defined attributes are simpler for it to match.

      The research acknowledges its limitations, noting it used a single scan per store, exclusively on Google Gemini rather than other platforms like ChatGPT or Perplexity. Additionally, the queries were algorithmically generated with categories assigned based on query language, resulting in 42% of stores remaining uncategorized. The stores were sourced from BuiltWith, and the queries aimed to align with each store's category, targeting questions shoppers ask when they know what they want but lack information on where to purchase it, rather than brand-name searches, which would be easier to dominate.

      “The shop window used to be Google. Now it is whatever the AI decides to recommend, and most brands have no idea whether they are included in the answer,” said Delian Coroamă, founder and CEO of Recomaze. “The stores that get recommended are not necessarily the largest; they are the ones whose product information can be effectively interpreted and trusted by an engine.”

      Recomaze has a vested interest in this issue, as it provides tools that monitor a store’s visibility across ChatGPT, Gemini, and Perplexity while also rewriting catalog data for better engine recommendations. However, the fundamental finding remains significant; a new intermediary now exists between shoppers and retailers, determining which names are recognized, and preliminary evidence suggests that many are overlooked.