Research · Category benchmark
Is 0% normal?
Every tool in this category will tell you your shop appears in 0% of AI answers, and let you assume that is your fault. We measured the shops around you. It mostly is not.
The answer is yes
shops appeared in none of the buying questions we ran in their own category. Every one of them scored above 90 for technical readiness — most scored a perfect 100.
We expected the category to explain the difference. It does not. The median appearance rate is zero in every single category we measured, and the strongest category still had most of its shops at nothing.
| Category | Shops appearing | Median | Best in category |
|---|---|---|---|
| Electronics | 4 of 12 | 0% | 100% |
| Pet supplies | 2 of 6 | 0% | 67% |
| Coffee | 2 of 7 | 0% | 17% |
| Fashion | 1 of 6 | 0% | 83% |
| Toys | 1 of 6 | 0% | 33% |
| Footwear | 1 of 7 | 0% | 17% |
| Natural beauty | 1 of 12 | 0% | 33% |
| Sports & outdoor | 1 of 13 | 0% | 33% |
The ordering between categories is not a finding, and we tested that rather than assuming it. The widest gap in the table above is 4 of 12 against 1 of 12; Fisher's exact test on that pair returns p = 0.32, which at these sample sizes is indistinguishable from chance. The consistency across categories is the result. The ranking is not.
An earlier draft of this page argued the opposite, built on a category that had four shops in it at the time. We measured eight more before publishing, and the contrast did not survive.
Why this is good news
An empty category is easier to enter than a crowded one. If most of your competitors were being recommended and you were not, you would be behind, and catching up is expensive. That is not the situation.
The situation is that assistants answer buying questions in these categories mostly without naming shops at all — and the first shops that become nameable take a position nobody is currently defending.
What the assistants read instead
An assistant that names no shop is not answering from nowhere. It reads pages. Across all 69 shops, the same handful of sources keeps coming back — and none of them are shops.
| Source | Times cited | How a shop gets in |
|---|---|---|
| en.wikipedia.org | 27 | Needs independent coverage first |
| trustpilot.com | 12 | You can list yourself — no gatekeeper |
| uk.trustpilot.com | 10 | You can list yourself — no gatekeeper |
| which.co.uk | 7 | A journalist decides |
| yelp.com | 7 | You can list yourself — no gatekeeper |
| consumeraffairs.com | 5 | A competitor published the page |
| youtube.com | 4 | Reviews in your category |
| quora.com | 4 | Real participation only |
One of these is a form you fill in yourself. A Trustpilot profile has no gatekeeper: you submit, you are in, and it gets cited across categories. A Wikipedia entry is the opposite — it needs independent coverage to exist first, and anyone selling you one is selling you a deletion.
That difference is the practical takeaway. The routes into these pages are not equally open, and the open ones are worth doing this week.
Method
For each shop we generate buying questions from the shop's own categories and market — the kind a shopper types, like "best coffee beans to buy online in the UK" — and run each question several times through an assistant with live web search. Appearing means the shop was named or linked in the answer. Models are not deterministic, so every figure is a sampled rate over repeated runs, never a yes or no from a single pass.
This measures one model on one set of dates, and each assistant has its own index and its own sources. The shops were selected for scoring high on technical readiness, which makes the result more striking rather than less — but it also means this is not a random sample of the web. We publish the counts so you can weigh them yourself.
How to cite: Goppa Research (2026). Is 0% normal? AI shopping visibility by retail category. https://trygoppa.com/study/categories — CC BY 4.0.
No shop is named. We publish the counts, the categories and the method; the identity of individual shops stays out of it.