Research · Category cut · UK natural beauty

Perfect scores, zero mentions

UK natural beauty retailers are the most technically prepared category we measured. It has not bought them a single mention in an AI answer.

Measured 7 August 2026 9 retailers scored 3 tested for appearance From the 2026 Q3 study

The category is technically excellent

100

is the median AI readability score in this category. Five of the nine retailers scored a flawless 100 out of 100. Seven scored 90 or above.

That is the best result of any category in the study, and it is not an accident. Seven of the nine run on Shopify, which publishes a UCP manifest and an llms.txt automatically — the platform does the technical work whether the merchant knows about it or not.

The five that scored a perfect 100:

greenpeople.co.uk 100 / 100
blommabeauty.com 100 / 100
mypure.co.uk 100 / 100
evolvebeauty.com 100 / 100
wildrisingskincare.com 100 / 100

Scores across the nine ranged from 43 to 100, with a mean of 88.8. Unlike most categories in the study, every site here responded to our crawler — none of the 28% non-response we saw overall.

And it appears nowhere

We took three of these nine retailers and measured something different: how often each one shows up when an AI assistant is asked real buying questions in this category — six questions per store, the kind a shopper actually types.

Retailer	Readability score	Questions	Appeared in
Store A	100	6	0%
Store B	97	6	0%
Store C	43	6	0%

Eighteen questions. Zero mentions. The store with a perfect technical score did exactly as well as the store scoring 43 — which is to say, neither existed as far as the assistant was concerned.

Three stores is not a sample. We are not claiming this generalises to UK natural beauty as a sector, and we would say so even if it flattered us. What three stores can do is kill a hypothesis: if a perfect readability score were sufficient to be mentioned, one of these would have been. It was not.

What the assistant read instead

This is the part worth the page. When we measured, we also captured which sources the assistant actually pulled from to answer those questions. For UK natural beauty, they were:

uk.trustpilot.com cited 3×
hotlineskin.substack.com cited 2×
britainreviews.co.uk cited 2×
barefaced.substack.com cited 1×

Two of the four are Substack newsletters. A personal newsletter about skincare was read and used by an AI assistant to answer a shopping question; five retailers with immaculate structured data were not. The other two are a review aggregator and a small comparison site.

None of these four are big publishers. There is no Vogue here, no Guardian, no Boots. The pages that got cited are exactly the kind of page a small brand can get into — or write.

And who appeared in the answers

The retailers the assistant did name, across those same questions:

vita-beauty.co.uk 3 answers
the-beauty-store-uk.myprosandcons.com 3 answers
sallybeauty.co.uk 2 answers
beautygoddess.co.uk 2 answers

What this suggests, stated as a suggestion. Being readable puts a retailer in the set an assistant can pick from. What appears to decide the pick is corroboration — third-party pages that discuss the brand. In this category the corroboration layer is Trustpilot and a handful of independent newsletters, which is a far more reachable target than it sounds. We measured which pages; we did not run an experiment proving that getting into them causes mentions. Nobody has.

If you run a store in this category

The unhelpful advice would be "improve your structured data". Your category already has a median score of 100 — for most of you there is nothing left to fix there, and fixing it again will not move anything.

What the data points at instead:

Reviews on the aggregator the assistant actually reads. Trustpilot was the single most-cited source here.
The small independent newsletters in your niche. Two of the four cited sources were Substacks. Getting covered by one is a realistic outreach target in a way that national press is not.
Comparison pages that include your competitors. The pages assistants cite compare. A page that only praises you does not answer the question a shopper asked, so it does not get used.

Method and limits

Nine retailers drawn from organic results of commercial searches in this category, with marketplaces and large chains excluded. Readability scored across 17 checks. Appearance measured on three of the nine, six questions each, on 7 August 2026.

Three stores is a probe, not a sample. Treat the appearance result as a reason to look further, not a category verdict.
One assistant, one day. Model answers vary between runs and change over time. These are not permanent facts.
We measure machine readability, not shop quality. Nothing here says anything about the products, prices or service of any business named. A score of 43 means a crawler struggled, not that the shop is bad.
We do not name the three stores measured for appearance. Publishing "this named retailer is invisible to AI" would harm a business that has done nothing wrong and does not know. The named retailers above are named for scoring 100, for being cited, or for appearing.

Check your own store free Read the full 2026 Q3 study

Goppa (2026). Perfect scores, zero mentions: UK natural beauty and AI assistants. Category cut from The State of AI Shopping Readiness, 2026 Q3. Retrieved from https://trygoppa.com/study/uk-natural-beauty