Ahrefs’ December 2025 analysis of 75,000 brands found stronger associations between AI answer visibility and brand mentions than with several backlink metrics. The study measures correlations, not the effect of changing one marketing activity. Use it to form a hypothesis, then test relevant changes against your own repeated measurements.
- YouTube mentions are the single strongest correlate with AI visibility across every platform: 0.737 on ChatGPT and AI Overviews, 0.717 on AI Mode (Ahrefs, 75,000 brands, December 2025).
- Branded web mentions rank second (0.656 to 0.709). ConvertMate and Digital Bloom put brand mentions at 0.664 versus 0.218 for backlinks, roughly 3× more predictive.
- Traditional SEO metrics underperform: domain rating tops out at 0.326, raw page count sits near 0.194, and backlinks stay very weak on all three surfaces.
- Google AI Mode correlates hardest with brand signals, acting as a consensus engine for established names. ChatGPT is the softer surface where challengers break in.
- Being named in an AI answer is not the same as having your site cited. Most brand mentions come from third-party pages, so earned coverage is the real battleground.
What correlates with AI visibility, ranked
Here is the order that matters, and it is not the order your SEO dashboard reports. Ahrefs measured how strongly each signal moves with brand visibility across ChatGPT, Google AI Mode, and AI Overviews. The pattern is consistent: signals of brand presence in the wider web dominate, and signals of technical authority trail. Parse tracks AI visibility across ChatGPT and Google AI Mode, analyzing 4.8 million AI responses and more than 57 million citation observations across 611,000 brands, and the same hierarchy shows up in which brands crowd the front of answers. Read the table below as the scorecard AI grades you on, then compare it to the one your current reporting optimizes for.
| Signal | Correlation with AI visibility | What it actually measures |
|---|---|---|
| YouTube mentions | 0.717 to 0.737 | Brand named in video titles, transcripts, descriptions |
| Branded web mentions | 0.656 to 0.709 | Times the brand is named across third-party pages |
| Branded search volume | 0.352 to 0.466 | Direct demand for the brand by name |
| Domain rating | 0.266 to 0.326 | Aggregate backlink authority |
| Number of site pages | ~0.194 | Raw content volume |
| Backlinks | very weak | Link equity, the classic SEO currency |
The takeaway is uncomfortable for anyone who spent a decade building link authority: the things that cost the most rank near the bottom.
Why YouTube mentions top every platform
YouTube leading the list surprises most marketers, until you see how AI models source answers. A "YouTube mention" is any time your brand appears in a video's title, transcript, or description, and models mine that corpus heavily because it is dense, conversational, and constantly refreshed. Semrush's 2026 index found Gemini draws on a small pool that includes YouTube alongside Wikipedia and Reddit, and AmICited has documented YouTube overtaking Reddit as the most-cited social platform in AI answers. The mechanism is straightforward: a review video, a comparison walkthrough, or a founder interview creates a transcript that names your brand in context an LLM can lift. Most brands treat YouTube as a top-of-funnel awareness channel and never think of it as citation infrastructure. That is the gap. The YouTube AI visibility strategy is less about production value and more about getting your brand named, accurately and often, inside content the models already read.
Branded web mentions beat backlinks by 3×
The second-strongest signal is the one traditional SEO systematically undercounts: how often your brand name appears across the web, with or without a link. Ahrefs put branded web mentions between 0.656 and 0.709. ConvertMate and Digital Bloom measured the same thing at 0.664 and set backlinks at 0.218, roughly 3× weaker. The distinction matters because SEO teams are trained to chase the link and ignore the unlinked mention. AI models do the opposite. They resolve your brand as an entity from the frequency and consistency of your name in third-party text, not from the anchor tags pointing at you. A brand named in fifty industry articles with zero backlinks will out-signal a brand with fifty backlinks and no editorial mentions. This is the clearest place where AI visibility and traditional SEO diverge: one counts links, the other counts language.
If you want to see how AI engines describe your own brand, run a free brand check. It takes a minute.
Branded search volume is the demand signal AI reads
Branded search volume correlates 0.352 to 0.466 with AI visibility, and Digital Bloom's earlier work called it the single strongest predictor at r=0.334. It is a proxy for something models care about: real, named demand for your brand. When people search your name directly, you accumulate the web presence, coverage, and entity clarity that AI answers reward, so branded search is both a cause and a symptom of brand strength. Digital Applied frames branded queries as a durable moat precisely because they are hard to fake and hard for competitors to erode. The practical read is that the top of your AI-visibility funnel is ordinary brand building: the awareness work that makes people search for you by name is the same work that makes models recall you by name. This is why branded and unbranded prompts split so sharply, and why demand generation quietly doubles as AI visibility work.
Why AI Mode rewards big brands more than ChatGPT
The three platforms do not weigh these signals equally, and the difference is strategic. Ahrefs found Google AI Mode correlates consistently harder with traditional brand signals than ChatGPT or AI Overviews, which makes it behave like a consensus engine: it favors brands with the deepest, most established footprint. ChatGPT shows weaker correlations with authority metrics, which Ahrefs reads as the best entry point for emerging brands. That is not a rounding error; it is a targeting decision. If you are a challenger, ChatGPT is the surface where sharp, well-sourced content can outrun an incumbent's brand mass. If you are the incumbent, AI Mode is where your accumulated brand strength pays off most. We covered the broader pattern in whether AI search favors big brands: the bias is real on some surfaces and beatable on others, and knowing which is which changes where you spend.
The platform split is a media-planning insight, not a trivia point. A challenger that spreads effort evenly across ChatGPT, AI Mode, and AI Overviews is fighting hardest on the surface (AI Mode) that rewards brand mass it does not have yet, and underinvesting in the one (ChatGPT) where content quality can win this quarter.
What barely moves the needle: backlinks, page count, ad spend
Domain rating, the headline number in most SEO tools, correlated only 0.266 to 0.326 with AI visibility. Raw page count sat near 0.194, meaning publishing more pages for its own sake does almost nothing. Backlinks, the currency SEO has optimized for two decades, stayed very weak across all three platforms. None of this means links are worthless; they still help classic organic rankings, and those rankings still feed some AI surfaces indirectly. But if your AI visibility strategy is a link-acquisition strategy with the label changed, the correlation data says you are working the wrong lever. The same goes for ad spend: nothing in the factor list rewards it directly. Models extract from earned language about your brand, not from your media plan.
Being mentioned is not being cited
One finding reframes the whole exercise. Semrush's 2026 index, built on 126 million prompts, stresses that being mentioned in an AI answer does not mean your own website is cited. The two are separate outcomes with separate mechanics. Idea Grove found 91% of AI brand mentions come from earned media rather than the brand's own site, and AirOps put roughly 85% of brand mentions on third-party pages. So the correlations point outward: YouTube, branded mentions, and branded search all live off your domain. This is why site-only optimization plateaus. You can perfect your own pages and still lose the answer to a competitor with more third-party presence. Which domains AI models cite most shows where that off-site coverage actually accrues, and earned media for AI citation is the acquisition playbook for it.
How to turn correlations into a plan
Correlation is not a to-do list, but it is a priority order. Start by measuring where you stand on the signals that matter, not the ones that are easy to report. Track your branded web mentions and YouTube presence against the specific competitors you lose AI answers to, per platform, because the surfaces disagree: BrightEdge found brand recommendations line up across engines only about a third of the time. Then invest where the correlation is highest and your gap is widest. For most brands that means deliberate third-party coverage, presence in the review and comparison content models read, and getting named accurately in video content, before another backlink campaign. Turn it into a number with share of model, and use Parse's citation gap analysis to see which sources feed a competitor and not you. The signals are earnable; the first step is measuring the right ones.
Correlation between YouTube mentions and AI visibility on ChatGPT and AI Overviews, the strongest factor measured.
Correlation between branded web mentions and AI visibility, roughly 3× stronger than backlinks at 0.218.
Correlation between raw site page count and AI visibility. Publishing more pages for its own sake barely helps.
Of AI brand mentions come from earned media, not the brand's own website.
See where you actually stand
The correlation data settles the strategy question: invest in the brand-strength signals AI reads, not the link metrics your dashboard defaults to. But the correlations are averages across 75,000 brands, and your gap is specific. Run your category's commercial prompts, compare your branded presence to the competitors who beat you in AI answers, and find the sources feeding them. Search your brand in Parse to see where you land across ChatGPT, Google AI Overviews, and Perplexity, and which signals are yours to earn next.