For an ecommerce brand, AI visibility is whether an assistant like ChatGPT names your store and your products when a shopper asks it what to buy, and increasingly whether a shopper can buy from you without leaving the chat at all. It is not the same problem as ranking a product page on Google, and the usual store-side tactics, more pages and more keywords, do very little for it. What moves it is being the kind of source an AI system trusts enough to repeat: independent reviews, earned coverage, and a product catalog a machine can actually read. Here is how it works and what to do about it.
Sources: Adobe's June 17, 2026 Brand Visibility newsroom release, and Adobe Analytics data reported by Digital Commerce 360 (August 19, 2026), both linked in the article.
How much shopping is actually moving through AI now?
Enough that the trend line is hard to dismiss, even though the absolute share is still small.
Adobe, whose analytics sit on a large slice of U.S. ecommerce, reported in a June 17, 2026 newsroom release that traffic to U.S. retail sites from AI sources surged 1,324% between October 2024 and May 2026. By July 2026, Adobe Analytics data reported by Digital Commerce 360 showed that AI-referred traffic was still growing 62% year over year, drawn from more than a trillion visits across over 200 of the largest online retailers. The more interesting part is quality, not volume: in that July data, AI-referred visits converted at a rate 60% higher than non-AI traffic, the eleventh straight month AI traffic had outperformed on conversion, and those visitors generated 53% more revenue per visit. A shopper who arrives from an AI answer tends to arrive further along in the decision.
The buying step is moving into the chat too. In its "Buy it in ChatGPT" announcement, OpenAI introduced Instant Checkout and an open Agentic Commerce Protocol, built with Stripe, that lets a shopper complete a purchase inside ChatGPT rather than clicking out to a store. It started with single-item purchases from U.S. sellers and was built to extend to more than a million merchants. When the recommendation and the checkout both happen inside the assistant, being the product it recommends is the whole game.
Why does AI visibility work differently for a store than for a search ranking?
Because the assistant is answering a question, not returning a list, and it builds that answer from sources it already trusts.
When someone searches "best wireless earbuds under 100 dollars," Google returns a page of links and the shopper does the comparing. When someone asks ChatGPT or Google's AI Mode the same thing, the assistant does the comparing and hands back a short, confident shortlist, often with two or three named products and a one-line reason for each. Your store is either in that shortlist or it is invisible, and there is no second page to scroll to.
That changes what the work is. A store page optimized for a keyword can rank without anyone else ever vouching for the product. An AI shortlist is assembled from what independent sources say, so the question stops being "is my page optimized" and becomes "when a neutral source writes about this category, does my product come up, and does it come up described the way I would want." This is the same mechanism behind generative engine optimization: the model repeats the consensus of its sources, so the real work is shaping that consensus.
What does an AI assistant actually read when it recommends a product?
Three things, roughly in this order: third-party reviews and roundups, your own structured product data, and whatever the model already absorbed in training.
The reviews and roundups matter most because they carry independent credibility. A product named in a respected publication's "best of" roundup, or reviewed in detail by a reviewer the model has seen cited before, is exactly the kind of source an assistant leans on when a shopper asks what to buy. Customer reviews feed the same machine: a consistent pattern of specific, credible reviews across the places an assistant reads gives it something concrete to repeat, where a thin or contradictory review profile gives it nothing to stand on.
Your own catalog is the second input, and it is the part most stores get wrong. If your product data, the title, the price, the specifications, the availability, is locked inside JavaScript that only renders for a human in a browser, a model fetching the page may see very little of it. Clean, readable product information in the raw HTML, accurate structured data, and a sensible stance on whether you let AI crawlers in at all decide whether the assistant can describe your product correctly or has to guess. Adobe's own framing of this is blunt: many retailers, it noted, still have AI visibility gaps that leave potential revenue on the table.
Do reviews and earned coverage really change what AI says about a store?
Yes, and this is the lever an ecommerce brand actually controls.
An assistant's answer is only as good as the sources behind it, so the most direct way to change the answer is to change the sources. That means getting the product into the roundups, gift guides, and category comparisons that cover your space, and earning the kind of independent writeups that a model treats as evidence rather than marketing. This is earned media, and it is disproportionately what AI assistants cite when they answer a buying question, because a journalist independently recommending a product reads very differently to a model than a brand recommending itself.
For a store, the practical version of this is less exotic than it sounds. A product that solves a specific problem, backed by a real result or a genuinely different design, is pitchable to the writers who cover that category, and the ecommerce press playbook covers how to make that pitch land. The payoff is double: the coverage reaches human shoppers when it publishes, and it keeps working afterward as a source the assistant reads the next time someone asks what to buy.
What should an ecommerce brand actually do first?
Start by finding out what the assistants already say about your category, then fix the two things you control.
Ask ChatGPT, Gemini, and Perplexity the real buying questions a shopper in your category would ask, and write down which products they name and why. If you are absent, you now know the shortlist you are trying to join. Then fix the catalog so a crawler can read your products cleanly, and start earning the independent coverage and reviews that get a product onto the shortlist in the first place. One thing not to do is treat this as a volume exercise: spinning up more thin product pages or syndicating the same release across hundreds of sites does almost nothing here, because none of it adds the independent credibility an assistant is actually weighing.
What is the honest caveat?
That the share of shopping running through AI is still small, and the numbers come with asterisks.
Adobe's figures are vendor-reported, drawn from its own analytics footprint rather than independently audited, and some independent analyses put AI's share of total ecommerce sessions well under one percent so far. A brand that drops everything to chase AI referrals today is over-rotating. The honest read is that the direction is clear and the traffic that does arrive is unusually valuable, so the sensible move is to do the work that compounds, readable product data and earned credibility, which happens to be the same work that helps with human discovery and search. You are not betting the business on AI shopping. You are making sure that when it does become a larger share, your store is already a source the assistants trust.
The direct answer, for anyone who read this far
How does AI visibility work for ecommerce brands? An assistant answers a shopper's buying question with a short, named shortlist built from the sources it trusts, so your store shows up by being one of those sources: independent reviews and earned coverage that name your product, plus a product catalog clean enough for a machine to read accurately. Keyword-stuffed pages and mass release syndication do not move it. Earned credibility and readable data do, and increasingly the purchase itself happens inside the chat, which makes being the recommended product the entire contest.
If you want to know what the assistants currently say about your category and where your store is missing from the shortlist, request a quotation and we will map it with you.
