Guide
The Questions Your Buyers Ask AI That Your Site Never Answers
A buyer types a real question into ChatGPT before they ever visit your site: which accounting software suits a two-person agency, say, or whether a given supplier ships to Ireland. The model answers with names. The company it names is rarely the one with the stronger product. It is the one whose site already contains a sentence that answers that exact question. Call this a buyer-intent question: the specific, ordinary thing a customer types right before they buy, not the broad topic a company writes blog posts about. Most sites cover the topic and skip the question. That gap is what decides who an AI assistant recommends and who it never mentions at all.
Why does a competitor’s page get cited instead of yours?
An AI assistant answering a buyer question is not ranking pages the way a search engine does. It is pulling a passage that already states the answer, and citing wherever that passage lives. If a competitor’s comparison page says outright that their tool “works well for teams of two to five”, and your site only says you serve “businesses of all sizes” somewhere in an about page, the model has nothing to lift from you. The competitor didn’t win on quality. They won because they’d already written the sentence the question needed. This is the exact pattern the check we run measures.
What kind of question actually decides this?
Not the ones companies plan for. A founder writes an FAQ about warranty terms and shipping windows because those are the questions support tickets bring in. The questions a buyer asks an AI assistant before purchase are different, and less predictable: “does this integrate with Xero”, “is there a UK-based alternative to [competitor]”, “can a sole trader use this without VAT registration”. These are narrow, situational, and usually phrased close to how the buyer would say them aloud. A site organised around product features rather than buyer situations answers almost none of them. Not because it’s dishonest. Because nobody wrote the sentence.
How do you find the questions your own site is missing?
You could type variations of your own category into ChatGPT, Claude and Perplexity by hand, and for a narrow category that’s a reasonable afternoon’s work. It gets harder once you have real competitors, because the questions worth checking are the ones your buyers actually ask, not the ones you’d think to ask about yourself, and a manual pass tends to catch only the obvious ones. This is a job we hand to an AI agent: it runs real buyer-intent questions for your category through the assistants, then returns the list of questions where your company never appears, with a note on where on your site to answer each one. What you’re actually looking for is not a score. It’s a list of sentences that don’t exist on your site yet.
Before any of that, it’s worth knowing what ChatGPT already says about your business today, because the gap usually starts from a specific, checkable answer rather than silence.
What should you do with the list once you have it?
Not write more content in general. That instinct produces another round of topic pages restating what your product does, which is exactly the kind of page an AI assistant already has plenty of and doesn’t need to cite. The fix is narrower and more mechanical: take the three or four questions your buyers already ask that your site currently answers nowhere, and publish a direct answer to each one, in a sentence that could stand alone if quoted. Three or four is enough. A comparison page that states plainly who a product suits, and who it doesn’t, gets cited more often than a features page written to sound universally appealing. If your business has never once shown up for a question a buyer is actually asking, that’s worth investigating on its own terms before assuming the fix is more content.
Common questions
Does adding an FAQ page fix this on its own? Only if the questions on it are the ones buyers actually type. An FAQ built from internal assumptions about what customers ask, rather than the real questions surfacing in AI answers, changes very little.
How is this different from normal keyword research? Keyword research optimises for what someone types into a search box to find a page. This is about what a model already answers without sending anyone to a page at all, which means the target is the sentence inside the answer, not a ranking position.
How do you find these specific questions? Run the actual buyer-intent questions for your category through the assistants, check whether and how your company is mentioned in each response, and keep the ones where you are missing, along with where on your site to answer them. Done by hand it is an afternoon; set up as a job for an agent, it repeats on a schedule.
Courage Horizon sets up AI agents for UK small businesses: software that does one job, such as chasing unpaid invoices, and asks you before it sends anything. Being found when customers ask ChatGPT is part of that work. Get a free report on your own business.