We put your buyers' real questions to ChatGPT, Claude, Gemini and Perplexity, then report, per question, with quotes, whether you're named, cited, or missing entirely.
15 questions · 4 engines · report in 24 hours
Named in the answerYour brand gets recognised.
Your site is citedA source buyers can actually follow.
8 questions × 4 engines · 32 answers
| Engine | Named / 8 | Cited / 8 |
|---|---|---|
| Gemini | 6 | 0 |
| ChatGPT | 7 | 6 |
| Perplexity | 5 | 1 |
| Claude | 4 | 0 |
Named in 22 of 32 answers, cited in just 7. Gemini and Claude say the company's name in most answers and never once send anyone to its site.
Real run, anonymised · a UK HR-software company · September 2026
A verdict per vertical.
Evidence for every claim.
They know your name. Your site is rarely the source they hand over. The fix is turning mentions into citations.
One rival owns 72% of answers here. Don't start this fight; there are cheaper ones waiting.
No vendor clears 25%. This is the cheapest question left to own, and we tell you which one it is.
We read your site and propose 15 buyer questions across 3 to 5 verticals. You edit and approve before anything runs.
All four assistants, asked from your market, web search on, every answer frozen the moment it lands.
Who's named, who's cited, which rivals got the recommendation. Each claim comes with its own quote.
A verdict per vertical and the one question worth winning first.
We check llms.txt, AI-crawler access, structured data, and what the bots actually see when they fetch your homepage. Thirty seconds, no account, nothing stored.
Readiness tells you whether you can be cited; the Snapshot tells you whether you are.
One-off or ongoing.
The evidence comes first.
A verdict, without a subscription.
See what changes each month.
The same evidence, for your clients.
Engines as of 14 September 2026: GPT-5, Claude Sonnet 5, Gemini 3.6 Flash, Perplexity Sonar — with web search on. Answers via each vendor’s API are not identical to the consumer apps; the methodology page says exactly how. Read the methodology ↗
London, GB · 2026-09-14 · 8 questions
London, GB · 2026-09-14 · 8 questions
London, GB · 2026-09-14 · 8 questions
Same models, via each vendor's API, with search on. Close, not identical. We publish the method so you can judge the gap yourself.
Because you need a verdict first, not a subscription. Track is for once you've started fixing things.
From your site and your verticals. You approve every single one before we spend a penny.
No. Every match is checked and printed, so you can see exactly what we counted and why.
Write the page that answers the question. We tell you which one. ScriptGrain can write it in your voice.