The set of AI questions
It is the core of the report: a fixed set of buying questions, in every language and market we work on, asked every month to ChatGPT, Perplexity and Gemini. The questions don't change from one month to the next. That way, if they start recommending you, it shows.
| For each question we record | Example |
|---|---|
| Who it recommends | The three options it names, in order |
| Which sites it cites as sources | Your site, a competitor's, a directory |
| Whether you appear and how | Recommended, only cited as a source, or absent |
| From which page | The exact URL the AI used |
- Search Console impressions and clicks, per language.
- New searches: the ones that show you and weren't in the baseline.
- Average position, read together with impressions.
- Indexing coverage of what gets published.
One indicator per stage
Each stage has the indicator that can be read at that point. We don't borrow the next stage's.
| Month | Leading indicator | Not measured yet |
|---|---|---|
| 1 | Frozen baseline, and the share of published pages Google indexes within 14 days | Clicks and positions |
| 2 | New searches that show you and weren't in the baseline | Conversion |
| 3 onwards | AI citations question by question, and an average position that holds while impressions grow | From here everything is read together |
What we don't promise
- A rise in clicks in the first month. If there is one, it is usually seasonality.
- Perplexity citations before month two, or Google AI summary citations before month three: the content has to be indexed first.
- Specific Google positions or specific AI answers. Nobody controls what a search engine says.