The State of AI Visibility in UK Uniform & Workwear

Five things the data shows

• “Which supplier” questions get a short list, every time. 99 to 100% of answers to our two supplier-list questions named at least one UK uniform supplier. The most-named supplier appeared in 70% of answers on one question and 59% on the other (Section 03).
• Advice questions name almost nobody. On five “how do I choose” style questions, 0 to 6% of answers named a uniform supplier. AI answered with fabrics, standards and garment brands instead (Section 04).
• AI says “I don’t know” far more often than it gets a fact wrong. Across 6,199 fact-checks, about 5% of answers were wrong and 35% gave no specific answer. Gemini gave no answer 65% of the time (Section 05).
• Reddit is a major source. Since 22 August, 12.6% of citations on the questions we track came from Reddit, more than twice LinkedIn, Trustpilot, Instagram, Facebook, YouTube and X combined (Section 06).
• AI almost never cites a negative source. 0.4% of 58,299 citations carried negative sentiment (Section 06).

Two question sets, not a market census

Every figure comes from buyer questions tracked in two SiteSignal accounts in this sector. One of them is The Uniform Consultants, a SiteSignal client. The questions were written for those accounts, so this is a close look at a handful of real buyer questions, not a survey of every UK uniform supplier. We only use neutral questions, ones that don’t name any company, because questions like “Is [company] a good choice…” tilt the results towards the company in the question.

One method change matters. Until 21 August 2026, our system recorded the websites an AI answer linked to or named. From 22 August it records the brand names written in the answer itself. The two measure different things, so we don’t compare numbers across that date. Our citation capture changed around the same time too (answers started logging about twice as many sources), so we don’t report citation trends either.

An answer means one question, asked once in one engine on one day. Engines: ChatGPT, Gemini and Perplexity through their APIs, plus Google AI Overviews where one appeared. Brand results use 22 August to 21 October 2026.

“Which supplier” questions: a short list, every time

Two question sets can’t rank a market of hundreds of suppliers, so this report doesn’t publish a supplier leaderboard. What the data does show is how concentrated the answers are.

The same handful of names came up day after day. What the most-named suppliers had in common on their websites: a page built for exactly the buyer in the question (corporate, bulk, multi-site), named clients or case studies, one page per sector, and clear delivery, returns and contact pages. We can’t prove any one page caused a mention, but the pattern was consistent.

“How do I choose” questions name almost nobody

If most of your content answers “how to choose a uniform”, it may help buyers learn, and AI may read it. It won’t get you named. Names appear when the buyer asks “which supplier”.

AI says “I don’t know” far more often than it gets a fact wrong

SiteSignal checks what AI engines say about a brand (name, founder, year founded, location, phone, email, policy links) against the brand’s own answers. In this sector we fact-check two brands, so these figures describe how the engines treated those two companies, not a rate for every supplier.

Two brands fact-checked in this sector, 18 May to 22 October 2026. Answers of “Unknown” count as no answer, and a policy link that differs only by “www.” or a trailing slash counts as right.

Gemini usually won’t commit to a specific fact. ChatGPT commits most and is rarely wrong when it does. Perplexity sits in between and gives the most “partly right” answers. The fix is the same for every engine: state your facts plainly on one page (who you are, where, since when, how to reach you) and keep them identical everywhere they appear.

Where the answers come from

Reddit matters in this niche. Because our citation capture changed in late August, we report Reddit’s current level rather than a trend.

AI almost never cites a source to say something bad about a supplier. The bigger risk is quieter: not being in the answer at all.

Can AI read your site at all?

On 15 October 2026 we checked the homepages of the 60 suppliers that came up most in our tracking. None blocked AI crawlers in robots.txt. 20 had an llms.txt file. 9 had no structured data on the homepage, and 9 didn’t load for our checker at all (a security page, or nothing). robots.txt isn’t the only gate: a firewall or bot-protection setting can turn AI crawlers away while robots.txt says they’re welcome.

Five things a uniform supplier can act on

• Ask AI the “which supplier” questions your buyers ask, in their words, not your brand name. Note who’s named and in what order.
• Build one clear page for each buyer you want: sector, size, problem. Add named clients and case studies.
• Don’t expect “how to choose” articles to get you named. They inform; supplier pages get named.
• Put your core facts on one page and keep them identical everywhere. AI often says “I don’t know” when facts are hard to find.
• Check your firewall lets AI crawlers in, not just your robots.txt. Then check again next month with the same questions.

Track your own category before assuming these numbers transfer to it. They describe a handful of real buyer questions in one niche.