Brand mentions are references to your company name across the web, with or without a link. In Ahrefs’ studies of 75,000 brands, they track visibility in AI answers much more closely than backlinks do. That finding changes where a GEO (Generative Engine Optimization) budget should go, but it proves less than most summaries claim.
- Ahrefs’ May 2025 study of 75,000 brands found branded web mentions correlated with Google AI Overview visibility at 0.664, against 0.218 for backlink count.
- Its December 2025 follow-up extended the analysis to ChatGPT and Google AI Mode, and found YouTube mentions the strongest signal at about 0.737.
- These are correlations on larger brands (Domain Rating above 40), not proof that mentions cause citations.
- Links still matter for classic rankings, so treat this as a reallocation of effort, not a reason to abandon link building.
What the Ahrefs studies measured
The claim worth testing is simple: if an AI system recommends a brand, something about that brand’s web footprint predicts it. Ahrefs tested several candidate signals against real AI answers, which is more useful than opinion.
In a study published May 26, 2025, Ahrefs analyzed 75,000 brands. It limited the sample to domains with a Domain Rating (DR, Ahrefs’ 0-100 link-authority score) above 40, took each brand’s top keyword with at least 800 monthly searches, and used Spearman correlation to compare web signals with brand appearances in Google AI Overviews. It tracked those appearances with its own Brand Radar tool.
A December 12, 2025 follow-up repeated the 75,000-brand analysis across ChatGPT, Google AI Mode and AI Overviews. For a buyer, the point is that two studies from one vendor point the same way, though the same vendor sells the tool that produced the data.
The numbers side by side
Spearman correlation: a measure from -1 to 1 of how consistently two rankings move together. A higher value means brands that rank high on one measure also tend to rank high on the other.
In the May 2025 AI Overviews study, Ahrefs reported 0.664 for branded web mentions, 0.527 for branded anchors (link text containing the brand name), 0.326 for Domain Rating and 0.218 for the number of backlinks. It also reported that about 26% of brands in the sample had zero AI Overview mentions. A brand with a modest link profile but frequent mentions therefore looked, statistically, more visible than one with the reverse.
| Signal | Correlation with AI visibility | Source and scope |
|---|---|---|
| Branded web mentions | 0.664 | Ahrefs, May 2025, AI Overviews, 75,000 brands |
| Branded anchors | 0.527 | Ahrefs, May 2025, AI Overviews |
| Domain Rating | 0.326 | Ahrefs, May 2025, AI Overviews |
| Number of backlinks | 0.218 | Ahrefs, May 2025, AI Overviews |
| YouTube mentions | About 0.737 | Ahrefs, December 2025, across ChatGPT, AI Mode and AI Overviews |
The December follow-up reported that link metrics showed very weak correlations across all three AI systems, while branded web mentions stayed strong in each one.
What the correlation does not show
Ahrefs states in both studies that correlation is not causation, and that deserves more weight than the headline numbers. A well-known brand attracts mentions and gets recommended by AI systems for the same underlying reason: people already talk about it. Mentions may be a symptom of popularity, not a lever you can pull.
The sample also skews large. Every brand had a DR above 40 and a keyword with real search volume, so the findings say little about a 10-person company with a narrow niche. We found no independent replication of these figures in this review, so treat them as one well-documented dataset, not settled law.
The implication for a decision-maker is to avoid two errors. Do not buy mentions, which would likely be low quality and is not what the data measured. And do not cancel link building, because the studies examined AI visibility, not Google rankings.
Where this shifts a GEO budget
If mentions are the better proxy for AI visibility, the budget question becomes where credible people discuss your category. Ahrefs’ own recommendations in the follow-up point at earning genuine mentions in articles and guides, building a YouTube presence, and avoiding low-quality links or content produced for its own sake.
That favors work a marketing team can defend: expert contributions to industry publications, original data other writers cite, partner and customer case studies, and video that explains your category. It also favors consistent naming. A brand described three different ways across the web is harder for any system to connect to one entity, which is also why structured data, covered in our guide to schema markup for AI search, supports the same goal.
A practical order of work
The sequence below starts with measurement, because acting on a correlation without a baseline leaves you unable to tell whether anything worked.
- Record a baseline: run your core buying questions through ChatGPT, Google AI Mode and AI Overviews, and note whether your brand appears. Our guide to measuring AI search visibility explains how to handle answers that change on every run.
- Audit where competitors who do appear are being discussed, such as publications, review sites and video channels.
- Pick two or three channels where your expertise is credible, and publish something worth citing there.
- Standardize how your company, product and founders are named across your site, profiles and press mentions.
- Re-run the baseline after a quarter and compare, keeping your link-building effort steady so you do not change two variables at once.
Frequently Asked Questions
Do backlinks still matter for AI search visibility?
They show a weaker relationship than mentions in Ahrefs’ data, but not zero. Backlinks also continue to influence classic Google rankings, which AI systems often draw on, so they remain part of a sound program.
Does a mention need a link to count?
Ahrefs measured branded web mentions as a signal separate from backlink counts. The studies do not establish whether an unlinked mention carries the same weight as a linked one in any particular system.
Can a small business expect the same results?
Not necessarily. The sample covered brands with a Domain Rating above 40, so smaller companies were not examined. The direction is a reasonable hypothesis to test against your own baseline, not a forecast.
Deciding where to earn mentions, and measuring whether it changes what AI systems say about you, is slow work that rewards a clear plan. At Grow with Hafsa we build AI marketing and search visibility systems around evidence like this, and we are glad to review your current footprint with you.