We didn't start with answers — we started with hypotheses. To understand what AI search is actually doing to web traffic, we pulled together a dataset of 100,000 unique websites, over 34 million Google Search Console queries, and 3 million LLM citations. Then we stress-tested the assumptions the industry has been treating as fact.
Most didn't hold up.
We tested many correlations that might be related to AI search. In this session, Bogdan Babyak walks through what we looked for, what we didn't find, and what the data actually revealed — including findings that reframe how you should think about AI visibility entirely.
This isn't a list of tips. It's a first-hand account of large-scale analysis: the dead ends, the surprises, and the signals that turned out to matter.