Original research · Published monthly
AI Citation Index
When someone asks an AI assistant which crypto exchange to use, or which wallet to trust, a handful of brands get named and the rest do not. We measure which ones — every week, across five answer surfaces, on a fixed set of the questions buyers actually ask.
Each month the results publish as a numbered edition. Every edition stays live at its own permanent URL, so the record is comparable over time rather than overwritten.
- 2 Sectors measured Exchanges and wallets
- 18 Brands tracked Ranked, not sampled
- 100 Buyer questions Fixed set, asked weekly
- 5 AI answer surfaces Named, not counted
Why it exists
Search used to send traffic. Increasingly it sends an answer.
An assistant that names three exchanges has made a shortlist on the buyer's behalf. Being absent from it is not a ranking problem that shows up in a rank tracker — it does not show up anywhere. This index is the measurement: who gets named, in which category, on which surface, and how that moves month over month.
Surfaces measured
- ChatGPT with web search
- Gemini with Search grounding
- Perplexity Sonar
- Google AI Overviews
- Google AI Mode
What gets ranked
Two sectors, each with its own set of 50 questions and its own nine contenders. Narrow on purpose: a sector asked 50 ways produces a number worth defending, where the same budget spread across six produces six that need caveats.
| Category | Brands | Questions |
|---|---|---|
| Exchanges | 9 | 50 |
| Wallets | 9 | 50 |
Editions
The first edition has not published yet.
Measurement runs weekly and an edition needs a full cycle of it before there is anything worth publishing. The scope above is fixed and the question sets are written; what is missing is the data, and the only way to get it is to wait for it.
Read the methodology · Including what this measurement cannot tell you