Key findings
AI-referred visits to US retail sites converted 90% better than other traffic in September 2026. Adobe.
51% of US consumers used an AI tool to support their shopping in the past month. NielsenIQ.
In ChatGPT beauty conversations, 54.7% of cited sources were retail and e-commerce domains. Similarweb.
How many shoppers ask AI before they buy?
Enough that it shows up in revenue. Adobe, which measures more than a trillion visits to US retail sites, reported that in September 2026 visits referred by AI assistants converted 90% better than other traffic and produced 83% more revenue per visit. In August 2025 the same traffic converted 9% worse. A shopper who arrives from an AI answer has already compared the options.
Use is broad as well. NielsenIQ found in September 2026 that 51% of US consumers had used at least one AI tool to support their shopping in the past month, a definition that includes retailers’ own recommendation features. In an Adobe survey of 5,000 US consumers in 2025, 38% had used generative AI for online shopping, mostly for research and product recommendations. Salesforce reported in July 2026 that agentic search as the first step of the shopping journey had grown 200% in a year.
What do the engines read to answer a product question?
More than the store. A product answer is assembled from several kinds of source, and the engine names the products those sources agree on.
Product pages and structured data. Price, specs, availability and policies.
Product feeds. The data an assistant uses to match a product to a question, and to check out.
Retailer and marketplace listings. Your product described by someone else, often from an older data sheet.
Review sites and buying guides. The roundups engines retrieve for best-of questions.
Reddit, YouTube and forums. Owners describing what the product is like to live with.
Similarweb found that retail questions are among the most likely to produce a cited answer in ChatGPT. In May 2026, 13.5% of retail conversations included web citations, against 6.8% across all conversations. In beauty conversations, 54.7% of the sources ChatGPT cited were retail and e-commerce domains. Other people’s listings carry your product’s facts.
What gets a product named in the answer?
Facts that match everywhere. A price, size or warranty that differs between your site, your feed and a retailer gives the engine a reason to name a product whose facts line up.
A page for the question, not only the product. Shoppers ask for the best option for a need and a budget. Comparison, sizing and care pages answer that directly.
Proof the engine already reads. Find which review sites, guides and threads are cited for your category, and work those first.
Data an assistant can act on. Google, Microsoft Copilot and Perplexity now offer checkout inside the assistant, and ChatGPT sends the shopper to the merchant’s own checkout. A product with incomplete feed data or unclear policies is harder to recommend and harder to buy.
What to do first, and sources
Write twenty questions your shoppers would ask, by category and price point. Run them logged out on ChatGPT, Gemini and Google. Note which products are named and which pages are cited. That list of pages is your work list. How SourceNode does this for retail and e-commerce. Every figure we publish, with its source, is on our numbers page.
The researcher
Odin Lund is co-founder and COO of SourceNode and leads its data analytics and research, including the daily tracking behind this work across ChatGPT, Gemini and Google AI Overviews. Odin founded First Query, the AI search practice that became SourceNode, and co-founded The Pixel Paper.
Meet the research team
About this research
Published by SourceNode
SourceNode is a data and research-first AI search firm. We record what AI models say about a business every day, trace each claim to its source, and use that data to get businesses named in AI answers. Every figure we publish carries its method and its sample.
Share Blog
Related research
AI search
AI search statistics for 2026: the numbers we trust, with sources
Read the research
AI search
What McKinsey’s AI search research says, and what to do about it
Read the research
