We type real searches into AI engines, write down exactly what comes back and publish the results in full. No surveys, no borrowed numbers. Every study below comes with its method, its limits and the complete results table.
Lower case, typos and all. We don't polish prompts for the engines.
Where an answer has a right or wrong, we check it against the official page on the same day.
Each study shows its full results table, so you can check our counting yourself.
Perplexity listed 12.3 sources per answer, ChatGPT 4.9 and Grok 3.8, though Grok said it read 44.5 pages on average. Only 24% of the sites Grok linked also appeared in Perplexity's sources for the same question.
You're welcome to quote, chart or build on any study here. Please credit Rank in AI Overview and link to the study page you used.
Writing about AI search and want a number we haven't published? Get in touch and tell us what you're looking for.
We run your brand through ChatGPT, Perplexity, Gemini, and Google AI Overviews by hand, then send you a plain-English readout of where you show up, where you don't, and the first three fixes — free.
No spam · no signup · one human reply with your audit
