Tools · 08 · built here
The Prompt-Distribution Builder.
A buyer never types your keyword into an AI engine. They ask a running sequence of natural-language questions — one per stage of the journey — and each one is a separate retrieval where you're cited or you're not. Enter what you sell; get the exact set of prompts you need to win, grouped by funnel stage.
Each stage is a different citation surface — a distinct question, retrieved on its own, where you are either named or you are not. Run every prompt through the LLM Visibility Audit to see which already cite you and which are still open.
Buyers ask; they don't keyword.
Nobody types a target keyword into ChatGPT. They ask real questions, one after another, as they move from noticing a problem to picking a vendor — and each question is a separate retrieval, a separate chance to be cited or omitted. You earn visibility query-by-query, not keyword-by-keyword. The structural work each citation depends on is documented in statement-level visibility.
Problem-aware
Feels the pain, hasn't named a category.
The buyer hasn't decided what to buy — only that something is wrong. A citation here frames the problem in your terms and puts you in the room before a category, or a competitor, is ever named.
Solution-aware
Knows the category, learning what good looks like.
They now know what kind of thing they need and are building a mental checklist. Cited here, you get to define the evaluation criteria the rest of the search runs against.
Vendor-comparison
Naming vendors, weighing them head to head.
The prompts start containing brand names. This is where a head-to-head citation decides whether you're inside the comparison or relegated to the footnote beneath it.
Decision / shortlist
Asking the model to recommend and rank.
The model is asked, outright, to name a winner or a shortlist. A citation here is the whole game — it is the difference between being on the list and never being surfaced at all.
Post-purchase / switching
Onboarding, retention, or eyeing the exit.
Migration and displacement questions. Cited here you catch buyers leaving a competitor — and you defend your own base from being talked out the door.
A prompt set is a hypothesis.
The set is a claim about where you ought to be cited. Running it is how you find out whether you are. Paste any prompt into ChatGPT or Perplexity and read the answer for your name — or run the whole set against fourteen models at once with the LLM Visibility Audit, then take the pages that lost to the GEO Diff to see exactly which structural dimensions to fix.
Run the set.
You have the questions. The LLM Visibility Audit runs these prompts against fourteen frontier models and shows where you're cited, where a competitor is, and where nobody is yet.
Run these prompts against 14 models →