For two decades, retail discovery has rewarded whatever already sold. A product with more buyers gets more visibility, more visibility brings more buyers, and the cycle repeats. Popular items get more popular while everything else stays buried.<\/p>\n
Shopify\u2019s<\/a> most recent quarter suggests artificial intelligence search is breaking that loop. Shopify President Harley Finkelstein said on the company\u2019s Aug. 5 earnings call that 75% of AI-attributed purchases on the platform came from outside Shopify\u2019s top 100 product categories, according to a transcript<\/a> of the call.<\/p>\n That same AI-referred traffic, and the orders behind it, each tripled year over year, PYMNTS reported<\/a>. Finkelstein pointed to specific examples surfaced through AI search that quarter, including a car seat sized for three-across sedan installation and reef-safe sunscreen that doesn\u2019t leave a white residue, according to the transcript. A keyword search buries both under higher-volume competitors. An AI system matching against a specific description can find them directly.<\/p>\n Retail Media\u2019s Whole Model Assumed a Human Was Scrolling<\/p>\n If AI systems keep taking over discovery, the advertising model retailers have spent two decades building starts to strain. Retail media networks have historically depended on human attention, with sponsored listings, search placement and visual promotions designed to persuade a person scrolling through a page.<\/p>\n Chris Selland<\/a>, founder of Differential Factor<\/a> and a lecturer at Northeastern University\u2019s D\u2019Amore-McKim School of Business<\/a>, sees that competition moving somewhere else entirely. \u201cAEO is the new battleground for digital shelf space,\u201d Selland told<\/a> PYMNTS, using the term for AI engine optimization. He described a future where product selection happens through an API call rather than a person browsing a page.<\/p>\n That shift changes what a retailer can even measure. Retail media today runs on impressions, clicks and conversions tied to a person\u2019s engagement. An AI agent reduces how much time a shopper spends browsing, so those numbers stop being a useful proxy for what actually influenced the purchase. Structured product data, fulfillment information, trust signals and availability become the inputs that determine whether an AI agent includes a retailer or product in its recommendation process at all, PYMNTS reported<\/a>.<\/p>\n Consumers Are Ready to Let AI Make the Call<\/p>\n Consumer demand is moving in step with that discovery shift. Nearly half of consumers, 48%, said they are at least somewhat interested in using AI agents to buy groceries or plan meals, according to PYMNTS Intelligence\u2019s \u201cThe Intelligent Spend Shift\u201d<\/a> report.<\/p>\n Consumers also expressed similar interest in letting AI agents manage subscriptions (48%) and handle gift purchases (44%), the report found, as commerce moves beyond simple recommendation engines. Those systems are increasingly capable of initiating a transaction on a consumer\u2019s behalf, not just suggesting what to buy.<\/p>\n For retailers carrying a large catalog, the consequence is concrete. A long tail of products never justified a homepage banner or a paid search budget, because there was no efficient way to connect that inventory to the right buyer. That changes once an AI system can find and describe a niche product directly to a shopper who asked for exactly that.<\/p>\n Promotions built around visual merchandising and urgency will not disappear, but the more agents mediate discovery, the more those promotions have to shape the preferences and rules an agent weighs rather than catch a shopper\u2019s eye in the moment.<\/p>\n