Research
Original data on how AI answers cite brands
We run a lot of prompts. Periodically we point that capacity at a question nobody has answered with data and publish what comes back — method, sample size and caveats included, whether or not the result flatters us.
What actually predicts brand mentions in Chinese AI search
Wikipedia and Baidu Baike presence predicted brand mention rate in Chinese AI answers. On-site schema density did not — and was mildly inverted.
72,000 answers
The Citation Volatility Index: how much AI answers change between identical prompts
We ran 400 identical prompts 25 times each across six AI answer engines. Citation sets changed on the majority of repeats, and the spread differs sharply by engine.
60,000 answers
Brands are named in AI answers 2.7x more often than they are linked
Citation counting undercounts brand presence badly. We measured the gap between being mentioned in the answer text and being linked in the source list.
90,000 answers
Domain concentration in AI answers: the top 100 domains take 58% of citations
AI answers concentrate on fewer sources than the classic SERP. We measured how tightly, and which domain categories are gaining share.
250,000 citations
How fast do AI answers pick up new pages? Median lag is 9 days
We published 120 controlled pages and watched six engines until they cited them. Median time to first citation was 9 days, with a long tail past 40.
43,200 answers