How do online shops get recommended by AI?
An AI assistant recommends a shop when the wider web — roundups, reviews, editorial and forums — describes it as a place to buy in your category, and when its crawlers can actually read your site. It is rarely about being the biggest retailer: in measured indexes, editorial presence beats audience size.
Updated 21 July 2026
Most writing about AI visibility is about big consumer brands. Shops are a different problem, and the advice does not transfer cleanly. This is the retailer version, from measuring one.
A retailer is not a brand, and assistants treat them differently
Almost every published "AI visibility index" ranks brands — the makers, the manufacturers. A buyer asking an assistant "where can I buy Cuban cigars online in the UK" is not asking for a maker. They are asking for a shop, and the answer names shops. That is a separate field with separate competitors, and being a well-known brand's stockist does not put you in it.
So the first thing to be clear about: you are competing to be named as a place to buy, against other places to buy — not against the products on your shelves.
Why the biggest shop often loses
The uncomfortable finding, seen across measured indexes, is that audience size barely predicts AI visibility. In one UK retail index the top brand won on editorial citations while a household name with twenty times the web traffic scored near the bottom. Small shops routinely out-appear large ones.
The reason is how assistants learn who belongs in an answer. They are drawing on what the web says about your category — "best online whisky shops UK" roundups, review sites, trade press, forum threads — far more than on how much traffic you get. A shop that reviewers and journalists mention is a shop assistants name. A big shop nobody writes about is not.
The two ways your shop gets named
An assistant surfaces a retailer one of two ways, and they need different work.
From memory (parametric). The model answers from what it absorbed in training. No page was fetched. You cannot edit that this quarter; the lever is to be widely and consistently described as a shop in your category across the sources training data is drawn from.
From live search (retrieved). The assistant searches the web mid-answer and writes from what it finds. This one moves in weeks: better pages, clearer buying guides, and citations from sources it trusts change your standing quickly.
In our own measurements most retailer mentions arrive from memory, with no citation attached — which means a plan built entirely on improving your own website is working on the smaller half of the problem.
The technical own-goal that makes shops invisible
Before any of that, a blunt check most retailers skip: can an assistant read your site at all?
We have measured shops that block AI crawlers outright, or whose pages need JavaScript to render, and then wonder why they never appear in AI answers. The two facts are the same fact. A site an assistant cannot read is a site it cannot cite, and no amount of reputation rescues a store the retrieval half of the system cannot open. This is checkable in minutes — our free scan reports it, and our crawler page shows what a shop should allow.
What actually moves it, in order
- Be readable. Server-rendered content, no crawler blocks, clean structured data. Cheap, fast, and it gates everything else.
- Be one unambiguous shop. The same name, address and description everywhere — your site, your listings, your profiles — so a model can tell your store from a similarly-named one.
- Get named in the category's roundups and reviews. "Best online [x] shop UK" articles, review sites, and the trade press for your sector. This is the biggest lever and the slowest, and it is where a small shop can genuinely out-punch a large one.
- Publish the buying answer, not the brochure. Guides that answer what a buyer asks — how to choose, what to look for, delivery and returns — are what retrieval lifts.
- Measure across providers, then re-measure. Being named by ChatGPT tells you little about Gemini. Anything else is a story.
How to find out where you actually stand
Not by asking an assistant about your shop once. Ask the questions a buyer asks — "where can I buy X online in the UK" — many times, across several assistants, and count how often you are named against the other shops that come up. That last part is the point: your visibility is your share of the shops named, not whether you appear at all.
When we did this for UK cigar retail, the honest reading for a real shop was 8 out of 100 — and the field we first measured turned out to account for under half the shops assistants actually named, which meant every share figure was overstated until we found the missing rivals. A dashboard watching one shop never has to look for them. An index does, because a share of voice is only as honest as its denominator.
You can check the readable half of your shop's standing with a free scan, and see how the retailers we measure compare in the index.