How it works
From your website to a fix list.
Four steps, then it runs on your plan’s schedule.
Steps
The loop
- 1
Add your brand
Enter your website. We suggest your category, competitors and buyer questions; you edit and confirm.
- 2
We ask the models
Every run asks ChatGPT, Gemini, Perplexity and Claude each prompt, 1–5 times per model depending on plan, so one lucky answer doesn’t skew the result.
- 3
We read every answer
Each answer becomes a ranked shortlist: who was recommended, in what position, the reasons given and the pages cited.
- 4
You get a brief and fixes
Your recommendation rate vs competitors, why you lose, and a ranked list of fixes. Ship one and track what changed.
Reliability
Samples and confidence
AI answers vary between asks. We ask more than once and show how sure each number is.
| Plan | Samples per prompt per model | Runs |
|---|---|---|
| Free trial | 1 sample | One-time |
| Starter | 3 samples | Weekly |
| Growth | 3 samples | Daily |
| Enterprise | 5 samples | Daily |
Every recommendation rate shows a 95% confidence range (Wilson interval) and a high / medium / low indicator. Each answer shows the exact model version the provider reported and the date it was asked. More samples, more days and more prompts narrow the range.
See where AI sends your buyers.
Run a free scan in about a minute. No account needed.