Working out how to appear in ChatGPT answers has quietly become a real commercial question for UK small businesses, not a novelty. Ofcom's Online Nation 2025 report recorded 1.8 billion UK visits to ChatGPT in the first eight months of 2025, up from 368 million in the same period of 2024. Some of those sessions end with a customer being told which plumber, accountant or salon to contact. This guide covers what actually influences those recommendations, and what a small UK firm can do about it without a big budget.
What decides whether ChatGPT mentions your business?
ChatGPT mentions a business when two conditions hold: the model or its live search can find you, and the information it finds is clear, consistent and corroborated. Mentions come from two distinct routes: knowledge baked into the model during training, and pages retrieved by live web search at the moment of the question. You influence the first through your footprint across the wider web, and the second through crawlable, well-structured pages that directly answer the question being asked.
The distinction between the two routes matters because they respond to different work. Trained-in knowledge is slow and cumulative. It reflects how often and how consistently your business appeared across directories, press, reviews and your own site before the model's training cut-off. Live search is fast and page-level. When a user's question triggers a web search, ChatGPT retrieves candidate pages, reads them, and cites the ones that answer cleanly. A page published last month can be cited in a searched answer, while it may take years to influence what the model "knows" unprompted.
For most UK small businesses the practical priority is the live search route, because it is measurable and responsive. The broader discipline of doing this deliberately goes by several names, unpacked in this comparison of AEO, GEO and LLMO for UK small businesses.
Does ChatGPT actually send customers to UK small businesses?
The audience is real and large. OpenAI's Sam Altman announced at DevDay in October 2025 that more than 800 million people use ChatGPT every week, as reported by TechCrunch. In the UK specifically, Ofcom recorded 1.8 billion visits to ChatGPT in the first eight months of 2025. Referral traffic per business is still modest compared with Google, but the visitors who do arrive tend to be late in their decision, because the assistant has already shortlisted them.
The 800 million weekly users figure comes from TechCrunch's report on OpenAI's DevDay announcement, and the UK visit numbers from Ofcom. Ofcom also found that 53 per cent of UK adults say they often see AI-generated summaries when searching, which suggests answer-style results are now the default experience rather than the exception. The honest caveat is that nobody publishes reliable per-business referral figures, so treat ChatGPT as an additional channel worth an afternoon a month, not a replacement for the fundamentals.
How do you let ChatGPT crawl your site?
Check your robots.txt file allows OpenAI's crawlers. According to OpenAI's own crawler documentation, OAI-SearchBot surfaces websites in ChatGPT's search features, GPTBot collects content that may be used for model training, and ChatGPT-User fetches pages when a user directly asks about them. If OAI-SearchBot is blocked, your site cannot appear in searched ChatGPT answers regardless of how good your content is, so this is the first thing to verify.
Visit yourdomain.co.uk/robots.txt and look for any Disallow rules affecting these user agents. OpenAI's bot documentation shows the exact syntax; to allow search visibility you want:
User-agent: OAI-SearchBot
Allow: /
Two common UK-specific traps. First, some hosting providers and CDN firewall presets block AI crawlers by default, so check your Cloudflare or hosting security settings, not just robots.txt. Second, some agencies blanket-blocked GPTBot in 2023 during the training-data debates and never revisited the decision. Blocking GPTBot is a legitimate copyright stance, but it trades away long-term visibility, so make it a conscious choice rather than an inherited one. A related, lightweight step is publishing an llms.txt file, covered in this llms.txt implementation guide for UK small businesses.
Which pages does ChatGPT prefer to cite?
ChatGPT cites pages that answer one specific question directly, near the top of the page, in plain language, with evidence. Pages structured as questions and answers, honest comparisons, pricing explanations and step-by-step guides get retrieved and quoted far more readily than generic service pages that open with three paragraphs about your passion for excellence.
Think about the assistant's job. It has retrieved eight pages and needs to compose one trustworthy answer in seconds. A page titled "How much does boiler installation cost in Leeds?" that opens with a direct answer, then breaks down the factors, is easy to quote. A page titled "Our Services" that mentions boilers in paragraph six is not.
The formats that consistently earn citations are answer-first FAQ pages, "cost of X in the UK" explainers, comparison pages that concede genuine trade-offs, and process guides with numbered steps. The same qualities drive citations in other assistants too, and the mechanics are similar to those in this guide to getting cited by ChatGPT and its companion on getting cited in Perplexity.
How do you make your business a clear entity?
An assistant can only recommend a business it can identify unambiguously. That means one canonical business name, address, phone number and service list, repeated identically across your website, Google Business Profile, Companies House record, directories and social profiles. Contradictions between sources make a cautious model drop you rather than guess.
Start with the boring audit. Is your trading name identical everywhere, or is it "Smith & Sons Ltd" on Companies House, "Smith and Sons Plumbing" on Google and "Smiths Plumbers Leeds" on Yell? Pick one public-facing form and align everything. Add Organization and LocalBusiness structured data to your site so the relationships between name, location, services and reviews are machine-readable rather than inferred.
Your Google Business Profile matters more than it first appears, because assistants and their search partners lean on it as a verified record of local businesses. If yours is unclaimed or half-complete, fix that before anything else, following this Google Business Profile setup and verification guide.
Do reviews and third-party mentions matter?
Yes, heavily. Assistants are trained to prefer corroborated claims, so a business described consistently by independent sources (review platforms, local press, trade directories and industry bodies) is safer to recommend than one that only describes itself. Your own website tells ChatGPT what you claim; third parties tell it what to believe.
For a UK small business the achievable wins are concrete. Membership pages on trade bodies such as Gas Safe Register, Checkatrade or your professional institute create authoritative pages that name you. Local journalism, even a quote in a regional business round-up, creates exactly the kind of independent mention models weigh. A steady flow of detailed Google reviews gives assistants both volume and specifics to draw on, and there are compliant ways to build that flow described in this guide to getting more Google reviews ethically.
Avoid the shortcuts. Fabricated reviews and paid link schemes are exactly the low-corroboration patterns these systems are built to discount, and they carry regulatory risk in the UK under consumer protection law.
How should you structure content around real customer questions?
Build pages around the questions customers actually ask, phrased the way they ask them, one question per page or section. Mine your inbox, call notes and quote requests for wording. Then answer directly, show your working with evidence, and include the UK-specific detail that generic national content omits: prices in pounds, regional considerations and relevant regulation.
This is where a small firm's unfair advantage lives. You know that customers ask whether a wet room adds value to a terraced house, or whether they need a waste carrier licence for garden clearance. National content farms do not. Each of those questions is a page, and each page should carry first-hand specifics: what you found on recent jobs, what surprised the customer, what the realistic pitfalls are.
Depth beats volume. Five pages that thoroughly answer five real questions will earn more assistant citations than fifty thin variations of the same service page. The broader playbook for making one strong page work across every AI surface sits in the AI visibility hub.
How do you measure whether it is working?
Track three things: referral sessions from chatgpt.com in your analytics, how often ChatGPT mentions you when you test your priority questions each month, and whether "heard about you from ChatGPT" starts appearing in enquiry sources. None of these is perfect, but together they show direction of travel within a quarter.
In Google Analytics 4, ChatGPT referrals arrive with chatgpt.com as the session source, so a simple exploration filtered to that source gives you the traffic picture. For mention tracking, keep a fixed list of ten questions a real customer would ask, run them monthly in a fresh chat, and record whether you are named, who is named instead, and which sources the answer cites. The competitors' citations are a ready-made target list of directories and formats to pursue.
Add a "how did you hear about us" field to your enquiry form if you have not already. Attribution from assistants is leaky. People read a recommendation, then Google you, so self-reported source data catches conversions your analytics will misattribute to brand search.
How long does it take, and what are the honest trade-offs?
Searched-answer visibility can move within weeks of publishing crawlable, well-structured pages, because ChatGPT's search reads the live web. Unprompted brand mentions from the model's trained knowledge take much longer and depend on years of consistent public footprint. The trade-off is opportunity cost: this work competes for the same hours as reviews, local SEO and paid channels, and it should not displace channels that already convert.
The sensible framing is that almost everything above (clear entity data, answer-first pages, corroborated claims, honest reviews) is also what current Google rankings reward, so very little of the effort is speculative. You are hardening one asset for multiple surfaces rather than betting on a single platform. The genuinely ChatGPT-specific work (crawler access, monthly mention testing, an llms.txt file) amounts to a few hours to set up and an hour or so a month to maintain.
Where firms go wrong is treating this as a campaign rather than plumbing. Assistant visibility compounds off the same foundations as everything else in your marketing, which is why it works best wired into one system that also captures the enquiry, follows it up and turns it into a review.
The short version of how to appear in ChatGPT answers: let OpenAI's crawlers in, become an unambiguous entity, publish direct answers to real questions with UK-specific evidence, earn independent corroboration, and measure monthly. None of it is exotic. It is the same disciplined clarity good businesses have always benefited from, now read by a new kind of audience with 800 million people asking it questions every week.
How to Appear in ChatGPT Answers (UK SMB Guide) — FAQ
How long does it take to appear in ChatGPT answers?
There is no fixed timeline, and anyone quoting one is guessing. ChatGPT's search feature pulls from a live web index, so a well-structured page that OAI-SearchBot can crawl may surface in searched answers within weeks of being indexed. Mentions that come from the model's trained knowledge move far more slowly, because they depend on your business appearing consistently across the wider web before a future training cut-off. In practice, UK small businesses should treat this as a compounding project. Publish genuinely useful answer-shaped pages, keep your name, address and services consistent everywhere, and accumulate third-party mentions. Expect early movement on searched queries first, with broader unprompted mentions following over a much longer horizon.
Should I allow OAI-SearchBot but block GPTBot?
You can, and it is a legitimate choice. OpenAI's crawler documentation describes OAI-SearchBot as the crawler that surfaces websites in ChatGPT's search features, while GPTBot crawls content that may be used to train its foundation models. Blocking GPTBot in robots.txt keeps your content out of future training data but does not remove you from ChatGPT search answers, which rely on OAI-SearchBot. The trade-off is that trained-in knowledge is one of the ways ChatGPT mentions businesses when users are not triggering a live search. If visibility is your priority, most UK small businesses are better off allowing both. If you sell proprietary content, such as paid research or courses, blocking GPTBot on those sections while allowing it on marketing pages is a sensible middle ground.
Do I need to pay OpenAI to appear in ChatGPT answers?
No. There is currently no paid inclusion scheme for organic ChatGPT answers, and no way to buy a recommendation in the way you can buy a Google Ads placement. ChatGPT selects sources based on crawl access, relevance, clarity and the credibility signals it finds across the web. That levels the field in one sense, because a well-documented independent plumber can be cited ahead of a national chain with a vague website. It also means there is no shortcut. The work is unglamorous: allow the crawlers, structure your pages around real questions, keep business details consistent, and earn mentions on directories, review platforms and local press. Treat any agency promising guaranteed ChatGPT placement for a fee with the same scepticism you would apply to guaranteed Google rankings.
Does optimising for ChatGPT also help my Google rankings?
Largely, yes, because the underlying signals overlap heavily. Pages that answer a specific question directly, cite evidence, load quickly and come from a clearly identified business tend to perform well in both places. Ofcom's Online Nation 2025 research found that 53 per cent of UK adults say they often see AI-generated summaries when searching, so Google itself is moving towards the same answer-synthesis model. The main differences are tactical. ChatGPT answers lean more on unambiguous entity information and third-party corroboration, while Google still weighs link authority and user behaviour signals more heavily. If you build pages for humans asking questions, with structured data and consistent business details, you are effectively optimising for both, and you avoid the trap of chasing one platform's quirks at the expense of the other.
Can a small UK business realistically compete with big brands in ChatGPT?
Yes, more realistically than in traditional search, particularly on specific and local queries. When someone asks ChatGPT for an emergency plumber in Stockport or a bookkeeper who understands CIS subcontractors, a national brand's generic service page is often a worse match than a local firm's detailed, honest page on exactly that topic. Assistants reward specificity and clarity rather than domain size alone. The realistic constraint is scope. You are unlikely to be the answer for broad national queries like best accounting software, and chasing those is wasted effort. Focus on the intersection of your services, your locations and the questions your actual customers ask, then make sure every claim on those pages is verifiable somewhere else on the web.
