SEEN RESEARCH
A cited video takes a source slot. There’s no sign it gets your brand named more.
We followed 70 beauty videos on YouTube from the day Google’s AI tools first cited them, to see whether the brands they name started turning up in more answers.
Change in how often the brand was named, in the three weeks after the first citation compared with the three weeks before, against the same brand’s other questions. The likely range is -1.0 to +0.6 points.
What we tracked
SEEN’s own tracking asks everyday questions about skincare, vegan luxury beauty and hair health in the US of Google AI Mode, AI Overviews and Gemini every day. Two household categories, cleaning and cooking, were added as a check. We found 70 YouTube videos that name a brand and that these tools started citing between late July and October 2026, on 627 questions across 95 brands.
How we tested it
For each video and question, we compared how often the brand was named in the three weeks before the first citation with the three weeks after, in each Google tool separately. Two comparisons take out what would have happened anyway: the same brand on its other questions over the same days, and videos cited on the same questions that don’t name the brand. We set out this test before looking at the results.
| Measure | Result |
|---|---|
| Videos, questions, brands | 70, 627, 95 |
| Brand named, three weeks before vs after | 23.3% → 22.8% |
| Change against the brand’s other questions | -0.3 pts (-1.0 to +0.6) |
| Videos that don’t name the brand | -0.7 pts |
| Gap between the two | +0.5 pts (-0.6 to +1.5) |
| AI Overviews only, first three weeks | +1.0 pts, gone by the second month |
| Upload to first citation | Median 4 days; 50 of 70 within two weeks |
No week in the three months after the first citation showed a rise. The small AI Overviews rise matched a short blip in a separate study on software brands, and it did not last.
What can we take from it?
The good part is real. A cited video takes one of the source slots in Google’s AI answer, new beauty videos get picked up within days, and some people may click through, though probably not many. What we found no evidence for is the video getting the brand named in the answer itself more often. That seems to come from the reviews, editorial lists and press that name the brand for the question being asked.
Keep in mind. This follows videos as they started being cited; it is not a controlled experiment. It covers three US beauty categories and three Google tools over about three and a half months. It can’t rule out effects smaller than about one point, or slower effects that build over many videos. Only five videos were single-brand reviews, too few to judge on their own. The upload-to-citation figure comes from videos newly cited in this period, so new uploads are common in the group by design. Brands, questions and raw records stay private.