AI referral traffic reorders the retail funnel
The path a shopper takes to a product is changing quickly, and the data now shows it. PYMNTS reported on July 20 that 48 percent of online shoppers used AI during their most recent purchase, and that AI-referred traffic to U.S. retail sites grew 393 percent year over year in the first quarter of 2026. Over two years, ChatGPT's share of product research climbed from 2 percent to 30 percent. Those figures describe a structural move in how discovery works. Shoppers increasingly open an assistant, ask a question and act on a synthesized answer, and retailers are seeing the referral pattern change accordingly.
We read this as a structural funnel reset. For a decade, search engine results and on-site reviews shaped the consideration stage. AI assistants now compress that stage into a single conversational answer, which means the inputs an assistant trusts determine which products surface. Retailers that mastered search optimization face a new optimization problem, one built around how models read reviews, forums and structured product data. The 393 percent jump in AI-referred traffic signals that the shift is already material to acquisition, and it will compound as assistant usage deepens.
Reddit positions its threads as the new review
Reddit has moved to claim a central place in this new discovery layer. Jen Wong, the company's chief operating officer, said that "reading honest, real-world experiences is now a more important factor in the purchase decision than reviews from professional critics or influencers." The company reports that 71 percent of surveyed Reddit users said they used the platform during the consideration phase of a purchase journey. That behavior gives Reddit a distinctive asset, a deep archive of candid discussion that AI models increasingly cite when they answer product questions.
The economics of that asset are already visible. Google pays Reddit roughly 60 million dollars a year under a data licensing deal, a sign that the forum's content has become valuable feedstock for AI answers. For retailers and brands, the implication is concrete. Conversations on Reddit now influence what an assistant recommends, which makes community sentiment a discovery input that feeds directly into acquisition. Monitoring and engaging in those conversations moves from public relations hygiene to a measurable factor in agent-led shopping.
Yelp turns 330 million reviews into an answer engine
Yelp is repositioning its own review corpus for the same reason. The company hosts 330 million local business reviews, and chief executive Jeremy Stoppelman described how AI changes their utility. "This chatbot can really understand 500 reviews in a second whereas a consumer might say, 'Well, I read the first five reviews, so I guess that's good enough,'" he said. An assistant that reads an entire review history produces a grounded recommendation, and the value of large, structured review datasets rises with it.
The strategic point for commerce-tech leaders is that review depth now compounds. When an assistant can digest hundreds of reviews instantly, the businesses with rich, recent and detailed review histories gain an advantage inside AI answers. Thin or stale review profiles will underperform in the same environment. We therefore treat review generation and review data quality as an AI-era discovery discipline, one that feeds directly into how assistants rank and describe a business or product.
Platform dependence becomes a strategic risk
This new discovery order carries a dependence risk that leaders should weigh carefully. Yelp relies on Google for more than 70 percent of its U.S. web traffic, a concentration that leaves its audience exposed to changes in how Google surfaces answers. As AI Mode and assistant experiences absorb more queries, intermediaries that sit between content and the shopper face pressure on their traffic and their business models. The same dynamic reaches retailers whose acquisition depends on a small number of platforms.
We see a clear lesson for commerce-tech strategy. Concentrated dependence on any single discovery gateway becomes fragile as assistants reroute traffic. Retailers should map how much of their demand now flows through AI intermediaries, and they should diversify the surfaces where their products and content appear. Building direct relationships and owned data assets reduces exposure to a platform's shifting logic. The 70 percent figure at Yelp is a warning about what over-reliance looks like when the discovery layer changes underneath a business.
Marketplaces and merchants face a discovery reset
The broader field is crowded and consolidating around this shift. Amazon, Google, Reddit, YouTube, DoorDash, Grubhub and vertical platforms such as Vagaro, ZocDoc and RepairPal all hold review and intent data that assistants can consume. Each is deciding how to license, protect or expose that data as AI answers scale. Model providers including OpenAI's ChatGPT, Microsoft and Mistral sit on the other side of the exchange, hungry for trustworthy, current signals about products and services. The result is a marketplace for discovery data that operated at a fraction of this scale two years ago.
For merchants, the reset changes where effort pays off. Structured product data, authentic reviews and active community presence feed the models that now steer shoppers. Investment in these inputs improves how an assistant represents a brand, which then shapes the recommendation a customer receives. We advise leaders to treat their product and review data as a supply chain into AI answers, with clear ownership, freshness targets and quality controls. The retailers that manage that supply chain deliberately will earn a stronger position in assistant-led shopping.
What commerce-tech leaders should do now
Three moves follow from this analysis. First, measure AI-referred traffic and its conversion, so the 393 percent industry trend becomes a metric the organization tracks. Second, invest in the inputs that assistants trust, which means structured product data, high-quality reviews and monitored community discussion on platforms like Reddit. Third, reduce concentration risk by diversifying discovery surfaces and strengthening owned channels, so a single platform's changes cannot cut off demand.
The recommendation economy rewards preparation. Assistants have already become a primary route to purchase for a large share of shoppers, and the data they consume determines which brands they name. Retailers that supply clean, current and credible signals will surface well inside those answers. The work is unglamorous and it is measurable, and it belongs to the commerce-tech function as much as to marketing. We expect the gap between prepared and unprepared merchants to widen through the rest of 2026.
