AI shopping traffic is growing fast and bringing shoppers who are more inclined to buy something along.
New data from Adobe Analytics shows that in May, traffic to U.S. retail websites referred by AI tools surged compared to the same month last year, and these visitors behaved differently than shoppers arriving through traditional channels like search engines or paid ads. Adobe’s findings suggest online shopping discovery is shifting toward a fragmented mix of pathways that includes AI systems. Moreover, the clients brought by AI solutions differ from typical customers discovering retail websites via simple browser search.
For years, ecommerce AI strategy meant optimizing for search engine rankings. A shopper typed a query into Google, compared results, and clicked through to a retailer’s site. AI-driven shopping works differently. A shopper might ask an AI assistant to compare products, explain features, or suggest options, and the AI narrows the choices before the person ever visits a retailer’s website. That means retailers may increasingly be competing for visibility inside an AI-generated answer, while search rankings everyone cared about in the past may no longer be as relevant. This is the essence of AI product discovery: much of the decision-making can happen before a click occurs.
According to Adobe Analytics, AI-referred traffic to U.S. retail sites rose 138% year over year in May. More notably, these AI-driven visitors converted 54% better than shoppers arriving from non-AI sources, i.e. a higher share of their visits ended in an actual purchase. Adobe also found these shoppers were 15% more engaged with retail sites, spent 53% longer per visit, and generated 53% more value per visit than average. These conversion rates suggest that AI-referred shoppers may already have a clearer sense of what they want by the time they arrive, and have stronger purchase intent rather than casual browsing.
However, Adobe also points to a complication: many retailers are receiving this AI traffic while large parts of their own websites remain difficult for AI systems to read and interpret correctly. This machine-readability gap could limit how well products are represented inside AI answers, even as AI search in ecommerce grows.
If AI shopping agents increasingly guide product selection, the question of who owns the customer relationship, whether it is the retailer, the marketplace, or the AI interface itself, becomes an open issue for the industry.
