Before a customer visits your website, calls your number, or reads a single review, there is a growing chance they have already formed an opinion about your business -- based entirely on what an AI told them.
ChatGPT, Perplexity, and Google AI Overviews are now answering questions like "who is the best SEO agency in Manchester", "what do customers say about [your business name]", and "is [your business] worth using?" directly in the search results. No click required. No visit to your site. The AI summarises your reputation, your services, your pricing, and your credibility in a single paragraph -- and most business owners have no idea what it says.
This is the AI search reputation problem, and it is more urgent than most businesses realise. According to a 2026 study by OmniEclipse that manually checked 1,700 businesses across 32 industries, 88% of businesses do not appear in ChatGPT results at all. Of those that do appear, a significant proportion are described inaccurately -- with wrong locations, outdated services, incorrect founding dates, or confused with a competitor of the same name.
At Think SEO, we have been conducting AI search reputation audits for UK businesses as part of its search services. This guide explains exactly what an AI search reputation audit is, why AI engines get businesses wrong, and the ten prompts every business owner should run right now.
An AI search reputation audit is a systematic process of testing what the major AI search engines -- ChatGPT, Perplexity, Google AI Overviews, and Gemini -- say about your business when asked by a real potential customer. It goes far beyond simply searching your own business name. It tests how AI describes you in competitive contexts, whether it recommends you unprompted, how accurately it represents your services and location, and whether it confuses you with a competitor.
The audit answers one core question: when someone asks AI about your industry, your location, or your specific business, does the AI help you win the customer or lose them?
This matters because AI search behaviour is fundamentally different from traditional search. In traditional search, a customer sees your listing and decides whether to click. In AI search, the AI makes a recommendation -- and the customer often acts on it without visiting your site at all. Research from Parse.gl's 2026 AI Search Statistics report found that 58.5% of US searches now end without any click to an external website. The AI has already answered the question. If that answer does not include your business, or describes it incorrectly, you have lost a customer who never knew you existed.
AI systems do not deliberately fabricate information about your business. Errors occur through predictable mechanisms that, once understood, can be systematically corrected.
Thin entity signals are the most common cause. When an AI model has very little data about your business from authoritative sources, it infers details based on patterns from similar businesses in the same industry. This inference can produce plausible but incorrect statements -- a wrong founding year, services you do not offer, or a location that belongs to a competitor.
Conflicting data across sources causes AI systems to hedge or guess. If your website says you are based in Manchester but an old directory listing from five years ago says London, the AI may state either -- or merge them into something that is accurate for neither.
Competitor confusion is more common than most businesses expect. Businesses with similar names, overlapping service areas, or identical industry categories can get merged in an AI model's understanding. A study by Pixelmojo found that when AI engines link your brand to the wrong same-named entity, you still get cited -- but the citation points at someone else's business.
Outdated training data means that even if everything was accurate at some point, AI models have a training cutoff. If your business changed name, rebranded, moved location, or pivoted its services before that cutoff, the AI may still describe the old version of your business.
Absence of corroborating signals is perhaps the most important factor. AI models gain confidence in information when multiple independent sources agree. A business that exists only on its own website, with no Clutch profile, no Trustpilot reviews, no Reddit mentions, and no press coverage, gives AI systems very little to work with. The AI either ignores it or guesses.
Based on our audits of UK businesses, the most common AI reputation errors fall into clear categories:
| Error Type | Example | Typical Cause |
|---|---|---|
| Wrong location | "Based in London" when you are in Manchester | Old directory listing or thin location signals |
| Wrong services | Lists services you stopped offering | Outdated website content or competitor confusion |
| Wrong founding year | States incorrect establishment date | Training data from an unreliable source |
| Missing entirely | Business not mentioned at all | No third-party corroboration signals |
| Competitor confusion | Describes a different business with a similar name | Identical names or overlapping entity signals |
| Wrong pricing | Quotes inaccurate price ranges | Industry average inference or outdated content |
| Weak description | Vague, generic summary that could apply to any business | Thin unique entity signals |
The "missing entirely" category is the most commercially damaging. According to the OmniEclipse 2026 report, 77% of businesses ranking on Google page one are invisible in ChatGPT. Google and ChatGPT use fundamentally different signals to select results. A strong Google ranking does not translate to AI visibility.
The following ten prompts form the core of a basic AI search reputation audit. Run each one in ChatGPT (with browsing enabled), Perplexity, and Google AI Overviews. Document every response with a screenshot.
Prompt 1 -- Direct brand search: "What is [your business name]?" This tests whether the AI knows your business exists and whether its basic description is accurate. Check: business type, location, founding year, founder name, services.
Prompt 2 -- Competitive discovery: "Who are the best [your service type] in [your city]?" This is the most commercially important prompt. It tests whether AI recommends you when a customer is actively looking for what you offer. Most businesses fail this test.
Prompt 3 -- Service accuracy: "Does [your business name] offer [specific service you provide]?" Tests whether AI accurately represents your service offering. Particularly important if you have added or removed services in the past three years.
Prompt 4 -- Location verification: "Where is [your business name] based?" Tests location accuracy. Particularly important for businesses that have moved, have multiple locations, or operate nationally from a specific base.
Prompt 5 -- Review and reputation summary: "What do customers say about [your business name]?" Tests how AI summarises your reputation. If you have no verified reviews on platforms AI trusts (Clutch, Trustpilot, Google), the AI may say it has no information -- or worse, infer from industry averages.
Prompt 6 -- Competitor comparison: "How does [your business name] compare to [main competitor]?" Tests whether AI can differentiate you from competitors and whether it accurately represents your relative strengths. Often reveals entity confusion issues.
Prompt 7 -- Pricing and value: "How much does [your business name] charge for [service]?" Tests pricing accuracy. AI frequently infers pricing from industry averages when no authoritative pricing information is available.
Prompt 8 -- Trust and credibility: "Is [your business name] legitimate / trustworthy?" Tests what trust signals AI associates with your business. Businesses with no third-party verification (Companies House, professional body memberships, verified reviews) often receive weak or hedged responses.
Prompt 9 -- Discovery without your name: "What [service type] companies in [your city] specialise in [your specific niche]?" Tests AI discovery without branded prompts. Research shows this is where most businesses are invisible -- they only appear when their name is already known to the searcher.
Prompt 10 -- Misinformation check: "What year was [your business name] founded and who runs it?" Tests factual accuracy on the two data points most commonly incorrect in AI responses: founding date and leadership.
A business with strong AI reputation signals receives consistent, accurate, and positive responses across all three major AI engines. When asked "who are the best SEO agencies in Manchester", ChatGPT cites it by name with a brief accurate description. When asked directly about the business, the AI provides the correct founding year, the correct founder name, the correct services, and the correct location -- without hedging or guessing.
This does not happen by accident. The businesses that appear consistently in AI search results have built what we call a strong entity profile: a consistent, corroborated, authoritative set of signals about who they are, what they do, and where they operate -- spread across multiple sources that AI systems trust.
The key sources AI systems trust most, based on citation analysis across thousands of AI responses, are: Wikipedia and Wikidata, Reddit, G2 and Clutch (for B2B services), Trustpilot, industry-specific directories, and editorial coverage in publications with high domain authority.
Your own website, no matter how well optimised, is the least-cited source in AI responses. A study of B2B brands found that 88% of citations in AI answers went to third-party pages, with only 5% going to the business's own domain. The AI recommends you, and someone else gets the click -- unless your entity profile is strong enough to be cited directly.
Once the audit reveals specific errors or gaps, the correction process follows a clear sequence.
Fix the source data first. Every AI error has a source. Identify the outdated directory listing, the conflicting address, or the missing information that is causing the AI to guess. Update or remove every outdated listing you can find, paying particular attention to Bing-indexed directories, as ChatGPT draws heavily from Bing.
Implement comprehensive entity schema. Add server-side Organization and Person JSON-LD schema to your homepage and about page. This gives AI crawlers a machine-readable source of truth for your business name, founding date, founder, location, services, and contact details. Critically, this schema must be injected server-side -- AI crawlers often do not execute JavaScript, so React-rendered schema is invisible to them.
Create an llms.txt file. This file, placed at the root of your website, is read directly by OpenAI's crawler (OAI-SearchBot). It should contain a factual, wiki-voice description of your business -- not marketing copy. Third-person, neutral, and structured like a Wikipedia entry. See our AI SEO services page for more on how we implement this.
Build corroborating third-party signals. This is the off-site work that most businesses skip -- and it is the most important. Claim and complete your Clutch profile. Collect verified Trustpilot reviews. Ensure your LinkedIn company page is complete and consistent with your website. Seek genuine mentions on Reddit in relevant communities. Consider a Wikidata entry if your business has sufficient history.
Use AI feedback mechanisms. Both ChatGPT and Google AI Overviews have feedback options. Use the thumbs-down button in ChatGPT to flag specific inaccuracies and explain the error. Use the feedback option beneath Google AI Overviews to flag incorrect information. These mechanisms are not guaranteed to produce immediate results, but they are worth using alongside your entity correction work.
The question is no longer whether AI search matters for your business -- it is how much damage a weak or inaccurate AI reputation is already costing you.
Consider the customer journey. A potential client in Manchester searches "best SEO agency Manchester" in ChatGPT. The AI recommends three agencies. Your business is not among them -- not because you are not good enough, but because your entity signals are weaker than your competitors'. That customer never visits your website. They never see your case studies, your pricing, or your testimonials. The AI made the decision for them.
Now consider the alternative scenario: the AI does mention your business, but describes you as "a general digital marketing agency" when you are a specialist AI SEO agency. Or it says you are based in London when you are in Manchester. Or it quotes a pricing range that is three years out of date. In each case, the AI is actively working against you.
An AI search reputation audit takes approximately two to three hours to conduct properly across all three major AI engines. The fixes -- schema implementation, llms.txt creation, directory cleanup, and third-party profile optimisation -- typically take one to two weeks to implement. The results compound over time as AI systems re-crawl and update their understanding of your business.
Think SEO conducts comprehensive AI search reputation audits for UK businesses. We test your business across ChatGPT, Perplexity, Google AI Overviews, and Gemini using the full 10-prompt framework above, identify every source of inaccurate or missing information, and provide a prioritised correction roadmap.
The audit covers:
Request your AI reputation audit and find out exactly what ChatGPT, Perplexity, and Google AI say about your business today.
Think SEO is a Manchester-based AI SEO agency led by Daniel Pattemore. The agency specialises in Generative Engine Optimisation (GEO), technical SEO, and AI search visibility for UK businesses.
Daniel Pattemore is the founder and SEO Director of Think SEO. With a focus on practical search engine optimisation, he specialises in AI SEO, Generative Engine Optimisation (GEO) and technical SEO.
Get a free SEO audit and discover exactly what's holding your site back. No obligation, no sales pressure.
Continue your AI search roadmap
Explore the next practical steps for measuring AI visibility, protecting your site from quality risks, and earning stronger visibility in AI-powered search.
Generative Engine Optimisation (GEO) is the practical work of making a business easier to find, understand and trust in AI-led search. Learn what it is, how it works and what to prioritise.
Google AI Overviews appear at the top of search results for millions of queries. This guide explains what triggers them, how to get your content cited, and what to do if your traffic has dropped.