Your buyers have stopped Googling. They ask ChatGPT and Gemini “what’s the best tool for…” and take the two or three names the answer gives them. Your audits show whether you can be cited. The Tracker shows whether you actually are — and every run hands back the evidence to act on.
That single answer is one of hundreds your buyers ask every week. Ask each of your real buyer questions across every engine, several times over, and the pattern stops being anecdotal — you can see exactly where you show up, where you’re only mentioned in passing, and where you’re simply not in the room.
Every hollow dot is a buyer question where the answer named someone else. That’s the map no analytics tool has been able to give you — until now.
Search gave you a position in a list. AI gives a paragraph — there’s no list to be tenth on. So instead of a made-up rank, each answer is recorded as three separate, checkable facts. Never blended into one number without showing the parts.
Your brand appeared in the answer. Visibility — but not necessarily a click.
Your domain was one of the answer’s sources. This is the one that sends a reader to you.
Named in the opening line, or buried at the end? Position is recorded too — plus which rivals appeared beside you.
These systems aren’t deterministic — the same question asked twice can name different companies. Any tool that screenshots one answer and calls it your visibility is showing you a coin toss. The Tracker asks each question several times and records every pass, so you see whether you show up reliably — and a one-off is never dressed up as a trend.
A single blended score would hide the one thing that tells you what to do: whether the engine even looked you up. The fix for “the live web can’t find you” is nothing like the fix for “the model never learned you exist.”
They search while answering. If they’re not finding you, the pages they can reach don’t say what they need. Fix: your pages.
In our pipeline these answer from durable third-party coverage — which moves slowly. A result here is worth more, and takes longer to earn.
Google builds these from pages it already cites. Earning a citation on those pages is the lever.
A single run — for example 30 prompts × 5 engines × 3 passes = 450 requested answers — hands back headline measurements, per-prompt verdicts, the competitor pages winning your lost prompts, the answers nobody owns yet, and a prioritized plan where every “why” quotes the run’s own measurements. All figures shown are illustrative.
Measurement without a next step is just anxiety. The Tracker sits at the centre of a loop: it names the exact pages losing you answers, you take them into an audit and rewrite, then you re-run.
Nothing happens automatically — the Tracker hands you the page, you decide the fix, you spend the credit. You stay in control the whole way round.
Anyone can claim their AI visibility improved. The Tracker shows it — because it re-runs the same questions and names exactly which ones flipped. Absent becomes cited. A guess becomes a receipt you can put in front of a client or a boss.
See which sites your whole category’s answers keep citing — including ones you never thought to check. It’s the shortcut to knowing where a guest post or a fixed listing will actually change what AI says.
Runs are credit-priced by size — prompts × engines × passes — with errored answers disclosed, excluded from your results, and refunded. Never silently zeroed. One credit balance covers the whole suite, and every account starts with 15 free credits.
Six: ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and Google AI Mode. They don’t all work the same way, so results are reported per engine rather than as one blended number.
Three things measured separately in every answer: whether your brand was mentioned, whether your domain was cited as a source, and where in the answer you appeared. Only answers actually retrieved count — failed calls are disclosed, excluded, and refunded, never counted as zeroes.
Because AI answers vary between runs. Multiple passes separate durable visibility from a one-off mention and show how consistent your presence really is.
No. Paste your own, or let Prompt Foundations suggest them from real search demand and buyer-journey stages.
Runs are credit-priced by size — prompts × engines × passes — with errored answers refunded. A full six-engine, 30-prompt run is 35 credits; smaller runs cost less. Every account starts with 15 free credits.
Free account, 15 credits, first run in minutes. See the answers your buyers see — then change them.