The strongest AI visibility signal
isn't a backlink. It's a mention.

The finding

YouTube mentions out-correlate every SEO metric.

Of every factor Ahrefs studied across a billion data points — backlinks, Domain Rating, page count, the whole conventional SEO toolkit — the one with the highest correlation to AI brand visibility was YouTube mentions, at 0.737. It held for both Google-owned products (AI Overviews, AI Mode, Gemini) and OpenAI's ChatGPT. No traditional link metric came close.

That's a reframe of the whole game. We've spent two decades treating the backlink as the currency of authority. In AI search, the data points somewhere else: the brands that get cited are the brands people are talking about — and video is where a lot of that talk is captured as text.

Why it likely works

Transcripts are training data, and mentions are trust.

Correlation isn't causation, so hold the mechanism loosely — but the candidates are reasonable. YouTube transcripts form a massive, openly accessible text corpus that models both train on and retrieve from. A brand named across dozens of reviews, tutorials, and comparison videos generates a dense, repeated, third-party signal that it exists and is worth recommending. And that signal is hard to fake at scale: it's people choosing to discuss you, which is precisely the "known, not optimized" pattern AI search rewards.

In other words, a YouTube mention is closer to a genuine reputation signal than a link is. Links can be bought and built; sustained video discussion is harder to manufacture, which may be exactly why the models lean on it.

Play 1

Earn mentions in the videos your buyers already watch.

Find the creators making review, tutorial, and "best of" videos in your category — the ones whose content already surfaces when you ask AI assistants about your space. Getting featured there can matter more than your own uploads, because it's third-party discussion. Offer access, data, a demo, or a genuinely useful angle. The goal is accurate, repeated mentions across credible channels, not a single sponsored spot.

Play 2

Publish where you have real expertise.

Owned video works when it's substantive: clear explainers, honest comparisons, "how to do X" walkthroughs that name your product in context. Each one adds a transcript that says, in plain language, what you do and who you're for — exactly the text a retrieval layer can lift. You don't need production polish; you need clarity and consistency. Say your brand name and your "best for" explicitly, because that's what ends up in the transcript.

Play 3

Write video for the transcript, not just the watch-time.

Most YouTube advice optimizes for retention and the algorithm. For AI visibility, optimize the spoken text. State the brand, the category, and the use case in full sentences ("Lynceus is an AI visibility tool for brands that want to see how ChatGPT describes them") rather than relying on on-screen graphics the transcript never captures. Caption accurately. The cleaner the transcript reads as standalone text, the more useful it is to a model.

The caveat

Treat it as a strong prior, then verify.

One study's correlation isn't a law of nature, and YouTube presence compounds with other signals — it generates discussion, links, and search interest — so isolating its effect is hard. The honest approach: build YouTube presence deliberately because it's one of the best-supported levers in the public data, then measure whether your AI citations actually move. Keep what works, drop what doesn't, and don't bet the whole strategy on a single coefficient.

The bigger point stands regardless of the exact number: AI search rewards being talked about. Backlinks were the proxy for that in the Google era. In AI search, the conversation itself — especially on video — is the signal. See why your Google rankings don't carry over.