First-Hand Content and AI Citations: What the Evidence Actually Shows

The only experiment points at the number, not the camera. The GEO study by Princeton and Georgia Tech (2024, around 10,000 queries) remains the only causal investigation: enriching content with statistics, expert quotes and source citations raised visibility in generative answers by 30 to 40 percent, and the effect is equalizing: low-authority pages gained the most, up to 115 percent at the top end. Keyword stuffing performed worse than no optimization at all.

The strongest correlational signals: mentions and video. In the Ahrefs analysis of 75,000 brands (2025), YouTube mentions correlated with AI visibility at r=0.737 and branded web mentions at r=0.664; classic backlinks only at r=0.218. Important: correlation is not causation; big brands get mentioned more anyway. But BrightEdge (2025) confirms the direction: YouTube's share of AI citations grew from 18.9 to roughly 39.2 percent within about a year, and around 60 percent of cited videos are formats that show and explain something.

Honest limit: Google officially states that AI Overviews and AI Mode require no special optimization and no special schema; structured data is hygiene, not a lever. The lever is the measurable experiment in front of the camera: tests, tolerances, numbers, named sources, plus distribution to places where third parties mention the brand.