Kevin Indig, author of Growth Memo and a veteran of 15 years in SEO and growth, has conducted extensive user-behavior studies on how everyday shoppers interact with AI-powered product recommendations in ChatGPT and similar platforms. His research, spanning five studies over roughly 24 months with 100+ participants and 400+ recorded tasks, reveals striking patterns in how consumers make purchasing decisions when an AI agent presents a shortlist.
The most significant finding is what Indig calls the shortlist effect: over 88% of participants accepted the AI's framing without question, and nearly 75% selected the item ranked at the top of the list. Only 3% of shoppers verified recommendations by searching elsewhere on the web, including publisher sites, affiliate platforms, or YouTube. The one exception to this top-heavy bias occurs when a shopper recognizes and trusts a brand; a recognized brand in position five can outperform the #1 recommendation. For commerce practitioners, this means that securing placement in an AI agent's top recommendation slot is now as critical as—or more critical than—ranking first in traditional search results, and that brand trust remains a powerful override to algorithmic positioning.
Beyond AI recommendations, Indig's research on traditional Google search found that over 96% of shopping queries now display product carousels, which reduce click-through on classic blue links by at least 50%. He also notes that AI engines typically run live searches across two to hundreds of related keywords before generating answers, making optimization for these fan-out queries essential for AI visibility. Indig emphasizes that Google's Merchant Center has become more informative than Search Console for understanding how products appear in AI results, positioning commerce as the beachhead for AI optimization strategy.