YouTube and AI visibility: what the Ahrefs 0.737 correlation actually shows

If you run marketing at an Indian brand, someone has probably sent you a slide in the last few months saying YouTube is the strongest driver of AI visibility, with 0.737 printed next to it.

The number is real. Ahrefs measured it across 75,000 brands and published it on 12 December 2025. The reading that most decks put on it is not what the study says, and the gap matters if you are about to move a content budget because of it.

I went back to the source. Three things get dropped between the study and the slide.

What Ahrefs measured

Ahrefs ran correlations between a set of brand and search metrics and how often a brand appeared in AI-generated answers, across ChatGPT, Google AI Mode, and Google AI Overviews. YouTube mentions, meaning the brand name appearing in a video title, transcript, or description, came out highest.

The published figures are ranges across the three platforms, not single values:

Source: Ahrefs, AI brand visibility correlations, Louise Linehan and Xibeijia Guan, 12 December 2025.

The shape of that table is the finding worth keeping. Every factor near the top describes something happening away from your website. Schema markup, heading structure, and llms.txt do not appear anywhere in the upper half.

Three things the headline drops

The sample is filtered to brands that have already won. The 75,000 are not 75,000 arbitrary companies. Ahrefs filtered to a domain rating above 40 and at least 800 monthly searches on the brand term. That is a screen for established demand. If your brand does not clear it, and most Indian mid-tier brands do not, the correlation is describing a population you are not in yet. This is the caveat I see dropped most often, and it is the one that decides whether the number applies to you at all.

Ahrefs say in writing that this is not causation. Their words: correlation is not causation, and spotting patterns between search metrics and AI mentions "doesn't mean improving these metrics will automatically boost your AI visibility." A study that disclaims the causal reading, quoted as proof of the causal reading, is how a correct number turns into a wrong decision. Brands that get cited by AI engines also tend to have heavy YouTube presence. Both of those can follow from being a large established brand without either one producing the other.

The study is from December 2025. A press release wave in late May 2026 recirculated it, which is why it keeps getting described as new. The measurement window sits before several engine changes in 2026, including how much weight Google's two AI surfaces now pull separately. That does not invalidate it. It does mean it is a starting position rather than a current reading.

Where our own data agrees, and where it stops

The part of the Ahrefs finding I have direct evidence for is the shape, not the channel.

Across 226 Indian brands in the June 2026 Cited Index, measured over ChatGPT, Perplexity, Gemini, Google AI Overviews, and Google AI Mode, 4.1% of brand citations point back to the brand's own website. The remaining 95.9% sit on domains the brand does not control: Amazon, Nykaa, Myntra, Reddit, YouTube, review platforms, industry editorial. We wrote that finding up in full in the 4% problem.

Owned share is not uniform. It moves with how the category buys:

A B2B software buyer ends up on product documentation, so the brand's own domain carries real weight. A parent choosing a baby-care product ends up on a marketplace listing or a review video, so it carries almost none. Same metric, and the allocation that follows from it is different in each row.

Here is where I have to stop short of the Ahrefs claim. Cited does not currently measure YouTube mention volume as a variable. We can tell you that YouTube appears in the third-party source mix for Indian categories. We cannot tell you that a brand raising its YouTube mention count sees its citation rate move, because we have not run that measurement. Anyone quoting our data to support the YouTube-specifically claim is going further than our dataset does.

That gap is worth naming rather than papering over. The honest position is that Ahrefs' global sample and our Indian sample agree that off-site presence is where citation volume lives, and only one of the two has anything to say about YouTube in particular.

What to do with the number

Treat 0.737 as a reason to test, not a reason to reallocate.

Start with your category's actual source mix rather than the study's average. If you sell into HR or CRM, a tenth of your citations come off your own documentation, and improving that documentation is measurable work with a known payoff. If you sell baby care or audio, that lever is worth about one citation in fifty and the argument for off-site work is much stronger.

Then pick the off-site channel your category's buyers actually use, and check it before committing. For Indian D2C, YouTube is a reasonable first candidate: India has the largest YouTube audience of any country, and most consumer brands already run video spend that is currently justified on awareness alone. Reddit is the other candidate worth checking, and we covered where it does and does not work in the Reddit play for Indian D2C.

Measure before and after on citation rate, not on video views. The whole point of the correlation caveat is that the two can move independently.

And keep the on-site work in proportion. It is a small share of citation volume in consumer categories, but it is also cheap, fast, and a precondition for anything else working. We laid out how we think about splitting the two in on-site versus off-site effort allocation.

If the Ahrefs relationship holds for Indian mid-tier brands the way it holds for their filtered global sample, a brand that doubles its YouTube mention count should see citation rate move within a quarter. We will run that measurement into a future edition of the Index and publish the result either way.

FAQ

Do YouTube mentions cause better AI visibility?

Ahrefs measured a correlation of approximately 0.737 across 75,000 brands, the highest of any factor they tested. They state directly that correlation is not causation and that improving these metrics will not automatically raise AI visibility. Brands AI engines cite tend to also have heavy YouTube presence. Whether one produces the other, or both follow from being an established brand, the study does not settle.

What was the sample?

75,000 brands filtered to a domain rating above 40 and at least 800 monthly searches on the brand term. That screen excludes most early and mid-stage brands, so the correlation describes companies that already have search demand and link authority.

When was the study published?

12 December 2025, by Louise Linehan and Xibeijia Guan, reviewed by Ryan Law. A press release wave in late May 2026 recirculated the findings, which is why it is often described as newer than it is.

How much of AI citation volume comes from a brand's own website?

4.1% across 226 Indian brands in the June 2026 Cited Index, measured over five engines. The other 95.9% sits on third-party domains. Owned share ranges from 1.7% in Baby Care to 10.0% in HR and Payroll.

Should Indian brands move budget into YouTube?

Not on this study alone. Off-site is clearly where the citation volume is, at 95.9% in our data. YouTube specifically is a candidate to test against your own category's source mix, because Cited does not currently measure YouTube mention volume and cannot confirm the channel-level claim for Indian brands.