Why your business is invisible to ChatGPT
AI engines answer with a handful of businesses. If yours isn't among them, the cause is usually one of four things — and three of them are fixable.
If a customer asks ChatGPT who to call for a blocked drain in Leeds, it answers with somebody. Not ten links — a shortlist, usually three to five names, often with a sentence explaining why each one is worth calling. If your business is not on that list, you are not ranking badly. You are absent.
The cause is almost always one of four things.
1. The engine cannot read your site
An answer engine is not a browser. Some crawl with their own agents, some rely on a search index, and several read a page without executing any JavaScript at all. A site that renders its content client-side can look perfect to you and arrive as an empty shell to the thing deciding who gets recommended.
This is the easiest of the four to check and the most common thing we find. Fetch your own homepage the way a crawler would — no JavaScript, no cookies — and read what comes back. If your services, your locations and your trading name are not in that response as text, no amount of content strategy will help, because there is no content to weigh.
The second half of this is permission. robots.txt decides who may crawl you, and the AI crawlers are not one group: OpenAI alone operates separate agents for training, for live search retrieval, and for user-triggered fetches. Blocking one and allowing another is a real and reasonable configuration. Blocking them all by accident, usually by copying a snippet from a forum post, is also real, and is silent — nothing tells you it happened.
2. Nobody else mentions you
Answer engines are heavily influenced by what other pages say about you. Not what your own site claims — what directories, local press, trade bodies, review sites and forums say. When an engine assembles a recommendation, it is drawing on sources it already trusts, and your own website is rarely the most-trusted source about you.
This is the uncomfortable one, because it is the one you cannot fix by editing your site. If the sources an engine trusts for "emergency plumber Leeds" are three directories and a local news roundup, and none of them lists you, the engine has nothing to recommend you from. The work is getting into those sources — which first requires knowing which ones they are.
That is measurable. When we run a check, we record every source cited in every answer, so the output is not "improve your authority" but a specific list of pages that were trusted ahead of you for your own service, in your own town.
3. Your location is ambiguous and the engine guessed wrong
This one is more common than it sounds, and it produces genuinely strange results.
Ask about "Birmingham" without qualification and an engine may reasonably answer about Birmingham, Alabama. We have measured this on our own corpus: a firm in Birmingham, Alabama appeared in the cited sources for a query explicitly about a UK city, and it was only caught because we check the country of every cited publisher. Manchester, Boston, Richmond, Windsor, Halifax and Hamilton all exist on both sides of the Atlantic.
If your trading area shares a name with an American city, you are competing for the engine's attention with businesses 4,000 miles away, and the engine may not have noticed the difference. Being explicit about your area on your own site — postcodes, county, the towns you actually cover, "UK" written out — is not SEO folklore here. It is how you disambiguate yourself in the source material.
4. You are cited, but never recommended
Being mentioned is not the same as being recommended, and the difference matters commercially. An engine may cite your page as a source for a fact about, say, boiler pressure, while recommending three competitors for the actual job. In a rank tracker this looks like a win. In your inbox it looks like nothing.
This is why the two are worth tracking apart. A check records whether you were mentioned and whether your site was cited as a source, and it keeps every source the engine trusted to build its answer — because those move independently, and that source list is what you actually have to get onto.
(We do not score “was this business recommended” as its own number. Pulling that reliably out of free text is harder than it looks, and we would rather report what we can defend than add a figure we cannot.)
What to do about it
In order of how much of the problem they usually explain:
- Confirm you are readable. Fetch your homepage without JavaScript. Read your
robots.txtand know which AI agents it allows. - Find out who is being recommended instead of you, for your service, in your area, on the engines your customers actually use.
- Get the sources that already rank into recommending you. Not all of them — the ones the engines cite for your service in your area.
- Measure again. Answers move. A single reading is a snapshot, and a snapshot cannot tell you whether something worked.
The one thing worth avoiding is optimising against a guess. Every claim in this article is either checkable on your own site in five minutes, or measurable. Anything a supplier tells you about AI visibility that cannot be traced back to a recorded answer is a supplier telling you a story.
Our free check records what ChatGPT, Claude, Gemini and Perplexity answer about your trade in your area, keeps the answers and the citations, and shows you both. It takes about a minute and needs no card.