Ahrefs Study of 75,000 Brands Finds YouTube Mentions Most Strongly Correlated With AI Search Visibility
A new analysis spanning 75,000 brands has quantified what many GEO practitioners suspected: YouTube brand mentions have the strongest statistical relationship with how often AI search systems recommend a brand, far outpacing traditional SEO signals like backlinks and domain authority. The research, published by Ahrefs through its Brand Radar product and synthesized in a widely-discussed analysis on September 3, 2026, provides the most comprehensive cross-platform correlation study of AI search visibility to date.
The findings carry immediate implications for how marketing teams allocate resources in an era where AI assistants — ChatGPT, Google AI Mode, AI Overviews, Perplexity, and Copilot — collectively process over 462 million prompts per month and increasingly serve as the first point of product discovery.
Background and context
The concept of "Generative Engine Optimization" (GEO) has been gaining traction since 2024, but the field has lacked the large-scale empirical data that traditional SEO has long relied on. Marketers have been operating on intuition and small-sample observations about what makes AI systems cite and recommend brands.
Ahrefs, which operates one of the largest web crawl databases and launched its Brand Radar product for tracking AI visibility, has been building the infrastructure to measure this systematically. Brand Radar tracks brand mentions across AI answers from six major platforms — AI Overviews (308.3 million monthly prompts), Gemini (31.5 million), Perplexity (31.4 million), ChatGPT (31.3 million), Copilot (30.9 million), and AI Mode (29.4 million) — giving it a dataset large enough to run meaningful correlation analyses.
The 75,000-brand study represents the most ambitious attempt yet to answer a single question: which distribution channels actually move the needle on AI search brand visibility? The answer, backed by Pearson correlation coefficients, challenges long-held assumptions about what matters in the age of AI search.
What changed — specific details
The study measured how each brand signal correlates with how frequently ChatGPT, Google's AI Mode, and AI Overviews mention that brand. The correlation scale runs from 0 (no relationship) to 1 (the two rise and fall together exactly).
Here are the key findings, ranked by correlation strength:
YouTube mentions: 0.737 — The single strongest predictor. When a brand is mentioned more frequently on YouTube, its AI search visibility rises proportionally. The correlation for YouTube view counts is nearly as strong at 0.717, meaning it is not just about having videos but about those videos being watched.
Critically, the study found that it makes no difference whose channel the mention appears on. A teardown from an independent reviewer, a comparison video, a sponsored segment, and the brand's own upload all count equally. The AI models appear to process YouTube mentions as brand signals regardless of source attribution.
Web brand mentions (with or without links): 0.656 to 0.709 — Simply being named on the web, whether or not that mention includes a hyperlink, shows a strong correlation with AI visibility. This is a significant finding: it means that podcast transcripts, newsletter mentions, roundup inclusions, and comparison posts all contribute to AI discoverability, even when they do not pass traditional link equity.
Anchor text links: 0.511 to 0.628 — When a brand name appears as the clickable text of a hyperlink, the correlation remains meaningful but weaker than unlinked mentions. This inverts the traditional SEO hierarchy, where linked mentions were always more valuable than unlinked ones.
Domain Rating: 0.266 to 0.326 — Ahrefs' own 0-to-100 score for link profile strength shows only a weak-to-moderate correlation with AI visibility. This is a dramatic demotion from traditional SEO, where domain authority was often the primary ranking predictor.
Backlinks and raw page count: close to zero — The raw count of backlinks and the number of pages a site has show almost no correlation with AI search visibility. The traditional link-building playbook, measured purely by volume, does not predict whether AI systems will recommend a brand.
Ad spend: 0.215 to 0.286 — Paid advertising investment shows a weak correlation with AI visibility, slightly above domain authority but far below content and mention-based signals.
Separately, the study confirmed that YouTube has become the most-cited individual site in Google's AI Overviews. YouTube now accounts for 5.6% of all AI Overview citations and grew 34% in six months, according to Ahrefs' tracking data.
Impact and implications
These findings should cause a fundamental reassessment of how marketing teams prioritize their channels for AI-era brand building.
For SaaS and software companies specifically, the implications are stark. The data shows that press coverage written about a company by third parties generates the lowest citation rates of any industry measured. According to Profound's analysis of 11.84 billion citations from April to July 2026, the median software company gets only 11.4% of its AI citations from third-party press coverage — the lowest of any industry tracked. Furthermore, no SaaS review site made ChatGPT's top 50 most-cited domains. Trustpilot is the closest, at position 35 with just 1.0% of citations.
For content strategists, the data suggests a clear hierarchy of effort:
- Invest in YouTube presence — not just your own channel, but getting your product mentioned in any YouTube content. The channel does not matter; the mention does.
- Pursue brand mentions across the web — podcasts, newsletters, community discussions, comparison posts. Whether they link to you is secondary to whether they name you.
- Maintain LinkedIn presence — Semrush's tracking of the top 25 domains across 13 weeks (230,000+ prompts, over 100 million citations) found LinkedIn was the one domain on a consistent upward trend across all three AI platforms measured. However, LinkedIn's share varies significantly: it represents 4.1% of ChatGPT citations in AnalyzeAI's dataset of 83,670 prompts, but 0.0% on both Claude and Perplexity.
- Traditional link building and domain authority matter far less for AI visibility than for traditional search rankings.
For SEO teams accustomed to measuring success through domain rating and backlink counts, the study delivers uncomfortable news. The signals that predict traditional ranking success — Domain Rating at 0.266-0.326, raw backlinks near zero — are almost irrelevant to whether AI systems recommend a brand. This does not mean traditional SEO is dead, but it does mean that the skillset required for AI visibility is materially different from the skillset required for organic ranking.
Historical precedent
The pattern of platform dependency is not new to those who have watched Reddit's role in AI citations. Reddit is the site ChatGPT cites more than any other, appearing in 16.7% of mentions as of July 2026 (Ahrefs Brand Radar data). When ChatGPT cites a social platform, Reddit accounts for 51% to 76% of those citations across all 14 countries measured by Profound.
But Reddit's dominance has collapsed twice. First, from roughly 60% to roughly 10% in August-September 2025, when Google dropped the num=100 search parameter that ChatGPT relied on to access Reddit content. Then again, from 3.8% to 0.5% in August 2026, after ChatGPT Search changed how it distributes queries (according to Promptwatch data).
The lesson is clear: a platform can change what this channel is worth to you overnight, and you find out after. The researchers explicitly warn against stacking those drops on top of each other — the two trackers are counting different things — but the pattern of sudden, platform-controlled devaluation repeats.
What has not been confirmed
Several important questions remain open:
- Causation vs. correlation: The study measures correlation, not causation. It is possible that brands with strong YouTube presence are simply better products or have stronger overall brand equity, and that the AI visibility follows from the underlying brand strength rather than from YouTube mentions specifically. The study design does not isolate YouTube as a causal driver.
- Which AI models specifically: The correlation coefficients are aggregated across ChatGPT, AI Mode, and AI Overviews. It is not clear whether the pattern holds equally for each platform or whether some AI systems weight YouTube content differently than others.
- Longevity of the pattern: AI models are updated frequently. The correlation between YouTube mentions and AI visibility could shift as models are retrained, as new data sources are integrated, or as AI companies change their retrieval strategies.
- Industry variation: The 75,000-brand study appears to focus primarily on software and SaaS brands. The correlation hierarchy may differ significantly for consumer goods, local businesses, healthcare, financial services, or other verticals.
What to watch next
The GEO measurement ecosystem is maturing rapidly. Ahrefs Brand Radar now tracks 462 million monthly prompts across six platforms. Tools like AnalyzeAI, Profound, and Promptwatch are building independent citation tracking datasets. As these tools mature, expect more granular data on:
- Platform-specific correlation patterns (does YouTube matter more for ChatGPT than for Perplexity?)
- Industry-specific benchmarks (the SaaS pattern may not generalize)
- Temporal stability (do these correlations hold after major model updates?)
- Intervention studies (if a brand increases its YouTube presence, does AI visibility actually rise?)
The Anypost analysis that synthesized this data promises to update its tier rankings as additional studies land. The methodology — reading the citation studies, checking sample sizes, and comparing correlation coefficients across platforms — provides a replicable framework for evaluating future GEO research.
Sources
- Anypost analysis of AI search channel effectiveness — Detailed synthesis of Ahrefs 75,000-brand study with correlation data, Semrush tracking, AnalyzeAI, Profound, and Promptwatch data
- Ahrefs Brand Radar — AI visibility tracking product covering 462M+ monthly prompts across six platforms
- Ahrefs — Official Ahrefs X/Twitter account, publisher of Brand Radar and the 75,000-brand correlation study
