· Practical intelligence from COR SolutionsAsk ChatGPT for a plumber, an accountant or a builder: does your business show up? We built tooling that captures the actual searches it fires and the sources it reads, so the answer is now measurable.
Ask ChatGPT to recommend a plumber in your town, or an accountant, or a builder. Does your business show up? Most owners have no idea. Search rankings you can check in an afternoon. AI answers felt like a black box. So we built a way to look inside it.
The thing nobody tells you about AI search is that it does not always search.
Ask ChatGPT a how-to question, like how to fix a common boiler fault, and it answers straight from its own training. No web search, no sources, nobody credited. Ask it "best boiler company in your town" and it does the opposite. It fires real web searches through its browsing tools, reads the results, and names specific firms. Those searches are real network requests, which is the whole reason you can watch what it does.
That split matters more than it sounds. A business can rank number one on Google for that how-to question, pull thousands of visitors a month from it, and still be completely absent from the AI answer to the same question, because there is no citation to win. Nobody gets credited when the model answers from memory.
For a lot of trades and service businesses, the "how do I..." guides are the biggest source of Google traffic. In AI search, that whole category is largely invisible. Not because the content is weak, but because the engine is not looking anything up to answer it.
The questions where ChatGPT searches are the commercial ones. "Who should I use for X." "Best company for Y near me." These are the questions with money behind them, and they are where being visible actually counts.
When it searches those, two things decide who gets named. First, directories. Checkatrade turns up more than any other single source, with Which? Trusted Trader and a handful of others behind it. Second, reviews. The firms that get recommended are almost always the ones sitting at 4.9 and 5.0 stars.
Run a spread of twenty buyer questions and a typical local firm shows up in only a few of them, usually the ones on its home turf. The firms that beat it are near-perfect on reviews. Directories plus ratings decide the shortlist.
ChatGPT is not the only one people ask. Perplexity answers the same questions, and the two do not lean on the same places. Both start with the trade directories, Checkatrade first for each. From there they split. ChatGPT stays on the established directories and review sites, Which?, MyBuilder, TrustATrader. Perplexity casts wider and pulls in community sources, Reddit threads and forums, that ChatGPT ignores completely. So being recommended by one is no guarantee you show up in the other, and the work to get visible differs by engine.
Watch the real traffic. The most honest method is to watch what ChatGPT actually does. We open a logged in session, ask a real buyer question, and capture the network calls behind the answer: the searches it fires, the pages it reads, and whether our client gets named. This is not a theory about how the model behaves. It is the real behaviour, captured off the wire, and it is the only way to see the exact searches it runs.
Automate it. Doing that by hand does not scale, so we wrapped it in a tool that runs the capture on a schedule and files the results. Now the question "did we show up this month" answers itself.
Use the API for scale. DataForSEO, one of the data providers we already use, has an endpoint that does the same job at volume. It is location locked, so you can ask "as if I were searching from the UK". It costs about half a penny a query, it needs no browser, and it covers ChatGPT, Perplexity and Gemini from one place.
We use both, and this is the important bit. The API gives you breadth: dozens of questions across every engine, cheaply, on a schedule. The live capture gives you the one thing the API leaves out, the actual searches the model fired. Neither replaces the other.
The headline is simple. You can now measure whether you show up in AI search the same way you measure Google rankings. It is no longer a shrug.
The levers overlap with local SEO more than you would expect. Reviews matter, directory presence matters, being the obvious trusted choice matters. But the behaviour is different, and the "answers from memory" gap changes where the opportunity is. Pouring effort into AI visibility for questions the model answers from memory is wasted. The commercial questions, where it actually searches, are where the work pays back.
This is early, and the engines change month to month. We are sharpening the tooling as they do. But for any business that wants to know whether it shows up when a customer asks an AI for a recommendation, the answer is now something you can measure, not something you guess.
Useful? See more of us on Google.
One click adds COR Intelligence to your preferred sources in Search and its AI features.
Read next
Preferred sources is the closest thing SEO has to a loyalty scheme: a reader picks you once, then sees you more often in Search, Top Stories and AI Overviews. Here is why that matters and how to try it, on our blog, right now.
The expandable question-and-answer dropdowns are gone from Google for every site. Stripping out your FAQ schema is the wrong response, and the reason why says something bigger about where search is going.
