Three out of every four AI-attributed purchases on Shopify came from outside the platform's top 100 product categories.
That’s the hard number Shopify President Harley Finkelstein presented on the company’s August 5 earnings call. It signals a fundamental shift in how online retail might work, according to a PYMNTS report.
For twenty years, digital retail was trapped in a popularity feedback loop. A product with more sales gets more visibility, which generates more sales, cementing it at the top of search results. Niche items, like a car seat designed for three-across sedan installation or a reef-safe sunscreen without white residue, two examples Finkelstein cited, stayed buried. Shopify’s data now suggests AI-powered search is breaking that cycle. The platform saw AI-referred traffic and the associated orders each triple year-over-year, with three-quarters of that new business ignoring popular, high-volume categories.
It’s not just more search. It’s a different kind of discovery, and it puts power in a new place. The traditional retail model rewarded what was already winning. AI discovery is beginning to reward what exactly matches a specific, sometimes obscure, need.
The Engine Shifting from Keywords to Understanding
The mechanics behind this shift are straightforward but profound. Legacy search relies on keyword matching and sales velocity. If you don't know the exact phrase for that oddly specific car seat, you'll never find it.
“A keyword search buries both under higher-volume competitors. An AI system matching against a specific description can find them directly.”
AI changes the query. Shoppers can describe a problem or a nuanced need in natural language. A system capable of semantic understanding can then match that query against comprehensive, structured product data, even if the keywords don't directly align. This converts the long tail of inventory, products that could never justify a major advertising spend because their audience was too niche, from dead weight into a monetizable asset.
It also means the metrics that have ruled retail media for a decade are becoming less relevant.
Retail Media’s Impressions Lose Their Power
Retail advertising networks were built on capturing human attention. Sponsored listings, promoted search placements, and homepage banners all assume a person is scrolling, browsing, and clicking. The entire business model runs on impressions, clicks, and conversion rates tied directly to that human engagement.
AI agents short-circuit that process. An AI shopping assistant reducing the need to browse makes those traditional engagement metrics a poor proxy for what actually influenced a purchase. As Northeastern University lecturer Chris Selland told PYMNTS, “AEO is the new battleground for digital shelf space,” using the acronym for AI engine optimization. He described a future where product selection happens through an API call rather than a person browsing a page.
The new inputs for success become structured product data, fulfillment reliability, trust signals (like reviews), and real-time availability. These are the factors that determine whether an AI agent includes a product in its recommendations at all. The competition moves from the visual storefront to a data layer invisible to the consumer.
This aligns with a broader move in commerce where the interface is receding in favor of automated, intelligent agents. It’s a process that echoes the shift happening in payments, where the financial function is becoming an embedded, invisible layer within other apps. For more on that operational shift, see our analysis Your App Becomes Your Bank in Embedded Finance Shift.
Consumers Are Primed for the AI Checkout Handoff
The demand side is already moving. According to PYMNTS Intelligence’s “The Intelligent Spend Shift” report, 48% of consumers are at least somewhat interested in using AI agents to buy groceries or plan meals. Similar interest levels exist for letting AI manage subscriptions (48%) and handle gift purchases (44%).
This isn’t about passive recommendation engines anymore. It’s about systems that can initiate transactions autonomously.
As AI begins to mediate discovery, retail promotions won't vanish, but their function will change. The goal will no longer be to catch a shopper’s eye in the moment of a scroll. Instead, promotions must shape the preferences, rules, and data parameters that an AI agent weighs when making its selections. It’s marketing to the machine.
“The more agents mediate discovery, the more those promotions have to shape the preferences and rules an agent weighs rather than catch a shopper’s eye in the moment.”
What’s Next for Builders and the Bottom Line
For the immediate future, the implications are concrete. Businesses with deep, specialized catalogs have a new, scalable channel to market. The value proposition of massive marketplaces, where discoverability favors the already-popular, faces a direct challenge from AI-curated platforms. We're already seeing fintech firms like Rakuten, which recently posted record $4.2B revenue after halting a long loss streak, succeed by targeting specific market needs rather than chasing universal dominance.
The necessary focus for any online seller now is AEO readiness. This means investing in rich, structured, and accurate product data feeds. It requires ensuring fulfillment is fast and reliable enough to meet the expectations an AI agent will set. Building authentic trust signals, like verified customer reviews and transparent policies, becomes a technical requirement, not just a branding exercise.
The core question is how quickly the discovery market will adapt. The old model is a powerful incumbent. But Shopify’s quarter shows the AI-attributed purchase metric is more than a vague promise. It’s a measurable, high-growth revenue stream skewed toward the niche. The feedback loop has a new input: specific intent, not just past popularity. If that signal holds, it redraws the map of who wins in online retail.
Why This Changes Everything
- It breaks a two-decade cycle where only popular products gained visibility, finally enabling niche products and small businesses to compete.
- AI-driven discovery shifts power from large-scale, high-volume retailers to producers of specialized goods that precisely match specific customer needs.
- This transformation could significantly increase e-commerce variety and consumer choice, boosting overall retail innovation and sales diversity.
Originally published on XOOMAR. For more news and analysis, visit XOOMAR.
