AI visibility for UK SaaS: the pound price, the VAT line and the British facts a buying question carries
Measured on 15 October 2026 across 204 UK readings of ChatGPT and Gemini, with one citation of our own domain: the British fields a software buying question contains, and the pages that answer them.
Published by AI Knows Us · Updated 16 October 2026
A British software company gets named by an assistant when the facts a UK buying question carries are in plain text on pages an engine can fetch: a price in pounds with the VAT position stated, a data protection position written under UK GDPR, a security certification with its certifying body and scope, a registered company number anyone can look up at Companies House, and a comparison page whose claims would hold up if the ASA read them. On 15 October 2026 we put 102 buying questions through ChatGPT and the same set through Gemini for the United Kingdom, 204 UK readings in total, and our own domain was cited once, on ChatGPT, and not at all on Gemini. The pattern in the UK set was not a preference for large companies. It was a preference for pages that were specifically British about a British question.
What the 15 October 2026 run measured, and what it did not
The run asked buying questions in one category, AI visibility, across three markets. No separate measurement was run for software, so nothing here is a citation rate for SaaS. What it gives a British software team is the source mix an assistant reaches for, and the UK market figure.
| Figure | Value |
|---|---|
| Date run | 15 October 2026 |
| Engines | ChatGPT and Gemini |
| Usable readings | 597 of 612 attempted |
| UK readings | 204, from 102 questions on each engine |
| UK readings citing our own domain | 1 on ChatGPT, 0 on Gemini |
| Citations recorded | 10,652 across 2,708 distinct domains |
| Vendor's or agency's own site | 85.6% |
| Official platform documentation | 9.7% |
| Review directories | 0.8% |
The 85.6% decides where a software team spends the quarter: across 597 readings, almost nine citations in ten went to somebody's own website rather than to a third party writing about them. Our product checks five surfaces, ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and this dataset covers two of them on one date. The method and the market comparison are in what AI assistants actually cite across 597 readings of ChatGPT and Gemini in three markets, and the British readings stand on their own in the UK cut of 204 readings.
What a British buying question contains that an American page leaves blank
Software buyers research in sentences, and a British sentence carries British constraints. An American page can be excellent and answer none of them, which is how a UK team loses a shortlist to a smaller competitor with a plainer site.
| What the question carries | The page that answers it |
|---|---|
| A budget in pounds, per seat or per licence | A pricing page with pound figures and the VAT treatment named |
| UK GDPR, and who the data controller is | A data protection page written to the UK regime, not only the EU one |
| A security question off a procurement form | A security page naming the certification, the certifying body, the scope and where data is held |
| Is this a real company | A registered name and company number, checkable at Companies House |
| Will this pass our accessibility check | An accessibility statement written against the Equality Act 2010 duty |
| How does it compare with the incumbent | A dated comparison page with the method shown |
Fix the pricing page first, in pounds
Every one of the 102 build spec conversations in the run asked for pricing in the reader's own currency, and that is the cheapest page to put right. Four things make a British pricing page quotable.
- Pound figures in the HTML. If the number arrives from a script after render, treat it as a number the fetcher may never see. View source and search for it.
- Say what the figure does with VAT. A price per seat excluding VAT and a price per seat including it are different answers to the same question, and an assistant repeats whichever one it finds. Write the treatment beside the figure.
- Name the billing unit in British terms. Per user per month, per licence, annual commitment, and what happens on renewal.
- Give the structure even where the price is bespoke. Bands, the unit, the minimum term, what moves the figure, and a review date. A page saying only to contact sales hands the answer to a competitor whose page has figures on it.
The security and data page is the one procurement actually reads
In the UK readings, gov.uk was cited 57 times and ico.org.uk 28 times, and neither domain appears in the US set. British official sources are already inside British answers, which tells a software company where its page belongs: beside the regulator's page as the practitioner layer, not against it.
So write the operator detail the regulator never will. Name the certification and the body that issued it, state the scope of the certificate rather than showing a badge, say which regions hold the data, list the sub-processors, and set out how a subject access request is handled. Where marketing consent is in scope, write it to UK GDPR and PECR, because that is the pair a British compliance reviewer checks against, and a page written to American rules reads as a page written for somebody else.
Write the comparison page as though the ASA will read it
Comparison and alternatives questions decide a shortlist, and they are where British advertising rules bite. UK marketing claims sit under the CAP Code and the ASA can be asked to look at them, so a superlative or a head to head claim on a British page needs its evidence attached. That is a constraint and an advantage at once: a dated comparison with the method shown is exactly the asset an engine will quote, and most competitors publish an undated grid of ticks instead. What the regulator can do is in what the ASA can do and what the CAP Code lets you claim.
A comparison page that holds up names the products compared, states the date, says how each row was established, and marks the rows where the competitor is stronger. The rows you concede are the reason the rest is believed.
Integration pages and docs do the quiet work
Integration questions are the most under served family in British software search, because the answer usually exists as a wall of logos and a logo is not text. One page per integration, naming the other product, what the connection does, what it needs and what it does not support, turns a graphic into pages that answer a real question. Documentation does the same job for how it works questions, and docs are usually the most quotable pages a software company owns.
Before any of that, check whether the fetchers can reach the site, the docs host and the help centre at all. The test is in whether AI search crawlers can read your website. The general version of this work, without the British grain, is in AI visibility for SaaS, and what changes between industries is in SEO by industry.
Questions about AI visibility for UK SaaS
Does a UK SaaS company need a .co.uk domain to be cited?
No. The British domains cited dozens of times in the UK readings were small, and what they carried was a page about the British version of the question. A .com with a UK pricing page, a UK data protection page and British wording carries the same signal. The targeting decision for ordinary search is in what actually differs for UK search.
Should the pricing page show figures excluding or including VAT?
For a business product, quote the figure the way your invoices quote it, and then say so in the same sentence. The failure is not the choice, it is the silence: a page with a bare figure and no VAT treatment gets repeated as though the treatment were obvious.
Do G2 and Capterra profiles decide a software answer?
They are worth keeping accurate, and they are a smaller lever than the industry assumes. Review directories took 0.8% of the 10,652 citations recorded on 15 October 2026, against 85.6% to vendors' own sites. Your own pages come first.
Is Cyber Essentials enough, or do UK buyers want SOC 2?
British procurement asks for the certification its own framework names, so publish what you hold, who certified it and what it covers, and let the buyer match it. Naming the body and the scope answers more procurement questions than naming a scheme alone.
How do I tell whether any of this moved?
Write the five question families your buyers use, with no vendor named in them, and ask each twice on ChatGPT and twice on Gemini, recording who was named and what was cited. Run it again in a month on the same wording. That is the whole measurement, and it is the one above with an N of 204 for the UK.
The British page that has not been written yet
The UK readings were taken by modest sites that had written something specifically British. Nothing in that result requires scale. It requires a pricing page in pounds, a data protection page written to the UK regime, a comparison that states its own method, and a date on all three. We sell this work, which is why the measurement here carries its date, its engines and its N, including the part where our own domain was cited once in 204 UK readings. AI Knows Us publishes that figure because it is the one we are judged on too, and a British software team can check its own version in an afternoon before buying anything.