The buyer prompt in one line

Every category has a sentence a real buyer would type when they do not already know your name. One concrete line. Not a keyword list. Not a page inventory. The question.

That line is where a Geo Audit starts. If you cannot write it, you cannot measure the answer. Engines return whatever the prompt asks for. Vague prompts produce noise. Precise prompts produce a shortlist you can argue with.

UK mid-market firms that already invest in being found usually start the other way round. They open the site map. They list service pages. They ask which URLs should be rewritten first. That is coherent work for rankings. It is the wrong start for generative naming. A model does not read your information architecture the way a crawler does. It answers a question. If the question is fuzzy, the shortlist is fuzzy, and every downstream finding is soft.

So write the category question a stranger would type. Name the buyer you actually want. Examples from accountancy and adjacent finance work: best accountant for a £10m UK firm; tax partner for an owner-managed manufacturer; FD support for a growing business in the South West. Those are examples, not a claim that every reader is an accountant. A law firm, a consultancy, or a specialist practice uses the same discipline with different nouns. The structure holds: buyer, job, constraint.

What fails is the vague line. “Accountants near me.” “Best firm.” “Who should I use” with no buyer and no job. The engine fills the gaps with whatever is easiest to cite. Your marketing partner then chases a ranking set that may not match the prompt a serious buyer ever types. Partners argue about screenshots that answered a different question than the one on the table.

One concrete line does more than set tone. We use it to build the rest of the locked set: brand prompts (does the model know you when named), category prompts (who it names when you are not), problem prompts (the pain in the buyer’s words), comparison prompts (you versus the firms that already hold the slot), and local or sector variants where they matter. The set stays fixed so the baseline can be re-run. Without that first line, the set has no spine.

The Audit does not start with a page inventory. It starts with the question. Pages matter later — as evidence a model can quote, or as gaps when a rival is easier to extract. First you need the prompt that decides whether you are in the conversation at all.

Partners often discover the real buyer sentence is narrower than the homepage claim. The firm sells “full service”; the buyer types a specific job and a size band. The firm talks “expertise”; the buyer types a place and a constraint. Writing the line forces that honesty early, before anyone spends two to four weeks measuring the wrong surface.

If you cannot write it in one line in a thirty-minute conversation, we should not take the brief yet. That is not gatekeeping for its own sake. A locked set built on a vague prompt wastes the fee and produces a named-instead list that does not match how your buyers actually ask. Better to stop, sharpen the sentence, and only then run the engines we can score.

Bring the line you believe a serious buyer types. We will pressure-test it, expand it into the set, and only then measure who is named instead of you. The answer is only as good as the question.