Answer capsule: Ahrefs studied 75,000 brands and found unlinked web mentions correlate with AI citations at 0.664, while backlinks correlate at only 0.218. That is a roughly 3x gap in favor of mentions. Brand search volume (0.334) and YouTube mentions (0.737) also outperform traditional backlink metrics. The implication is not that links are dead. It is that the marginal dollar of SEO budget buys more AI visibility when it is spent on earned conversations than on link placements.
Your SEO agency is telling you that you need more backlinks. Your competitors are getting recommended by ChatGPT and you are not. Both of these things can be true, and the reason sits in two data sets published by Ahrefs across 2025: the August 2025 analysis of 75,000 brands and AI Overviews and the December 2025 follow-up extending the study to ChatGPT and Google AI Mode. Taken together, they describe an off-page world where the things traditional SEO optimized for have a weaker statistical relationship with AI citations than the things traditional SEO largely ignored.
Soar is a community marketing agency that has run 4,200+ community campaigns across 280+ brands since 2017. This article is for the marketing leader who has to explain to her CFO why the 2026 off-page budget should not look like the 2022 off-page budget, and who wants the data and the plan in one place.
What the Ahrefs 75,000-brand study actually found
Answer capsule: Ahrefs measured Spearman correlations between brand properties and presence in AI Overviews, ChatGPT, and Google AI Mode across 75,000 brands. Unlinked brand web mentions led at 0.664. Branded anchor text followed at 0.527. Brand search volume came in at 0.334. Total backlinks came in at 0.218 - the weakest of the named off-page factors.
The headline number from the Ahrefs study is the 3x multiple between unlinked mentions and backlinks. But the rank order of the other factors is just as important for planning. Branded anchor text - which requires a link but also requires the brand to be named in the surrounding copy - outperforms generic backlinks by more than 2x. Brand search volume, which is a downstream proxy for "are people looking for this company by name?", also outperforms link count. The December 2025 follow-up extended the analysis across ChatGPT, AI Mode, and AI Overviews and reached the same conclusion: the top three correlations with AI brand visibility are all off-site presence factors, and links-as-volume sits below them all.
The extended study found something marketing leaders should read twice: YouTube mentions correlate at ~0.737 with AI visibility, the single strongest individual signal Ahrefs measured. YouTube transcripts feed both Google's and OpenAI's training pipelines.
So what for your brand: Treat backlinks as one signal among many, not the signal. The off-page budget in 2026 should look like a portfolio of mention-producing activity, not a pipeline of link placements.
Why mentions outperform backlinks for AI citations
Answer capsule: Backlinks were engineered to measure endorsement in a document graph that search engines crawl. Large language models learn from a different substrate - conversations, transcripts, long-form answers, reviews - where what matters is that a brand is named in context by a credible third party, whether or not the mention contains a hyperlink. Mentions are a better proxy for what LLMs actually ingest.
AI models are trained on mixtures of web text, licensed corpora, and conversational data. The Columbia Journalism Review reported that Reddit's data licensing business generated roughly $140M in 2025 alone, with OpenAI, Google, and others paying for access. Reddit conversations now represent a disproportionate share of the discussion-oriented text LLMs see. When those conversations name your brand in a context like "I use X for Y because Z," the model learns a statistical association between your brand and the category - whether or not the thread links out to your homepage.
The same logic applies to YouTube transcripts, Quora answers, G2 reviews, and long-form press. Per the Ahrefs December 2025 data, YouTube mentions correlate at ~0.737 with AI visibility because Google trained its AI surfaces on YouTube transcripts and OpenAI's GPT-4 was trained on more than a million hours of YouTube transcription. The link is not the carrier of the signal. The named mention inside credible discussion is.
So what for your brand: Campaigns that produce named brand mentions in community conversations, review platforms, and video transcripts will compound in AI answers. Campaigns that produce anonymous link placements on thin content pages will not.
How the off-page signals rank in 2026
Answer capsule: The table below summarizes the strongest off-page correlations with AI visibility from the Ahrefs 75K-brand studies, plus platform-level citation patterns from Semrush and Profound. Use it to reprioritize spend between link-building, mention-building, review presence, and community marketing in your 2026 plan.
| Signal | Correlation with AI visibility | What it actually is | 2026 budget priority |
|---|---|---|---|
| YouTube mentions | ~0.737 | Brand name in video titles, descriptions, and transcripts | High - creator and PR layer |
| Unlinked brand web mentions | 0.664 | Named mentions across news, forums, Reddit, Quora, reviews | High - community and PR layer |
| Branded anchor text | 0.527 | Inbound links whose anchor is the brand name | Medium - tied to digital PR |
| Brand search volume | 0.334 | Search demand for the brand's own name | Medium - downstream of awareness |
| Total backlinks | 0.218 | Link count from third-party domains | Lower - still a baseline |
| Press release syndication only | 0.04 (share of citations) | Wire distribution without editorial pickup | Low - near-zero citation share |
Sources for the table: Ahrefs brand visibility correlations, December 2025; Semrush most-cited domains, 2025; Wellows brand mentions vs citation analysis.
The key takeaway for budget planning: nearly every signal above the 0.3 correlation line is produced by activity that does not live on your website. Off-page work outside classic link-building is now the majority of the AI visibility lever.
Do backlinks still matter?
Answer capsule: Yes, but the job description has changed. Backlinks still move organic rankings, still contribute to domain trust, and still correlate weakly-but-positively with AI citations. What has changed is the opportunity cost. Every dollar that buys a generic guest-post link is a dollar not buying a branded mention, a review, a podcast feature, or a community thread - each of which carries a stronger signal.
The Princeton GEO study (Aggarwal et al.) found that citing external sources inside content improved a page's visibility in generative engines by 115% for lower-ranked pages, and that adding statistics improved it by 41%. Those findings validate on-page optimization - different from backlinks - and they do not override the correlation data. In practice, the brands we see winning in AI search run a smaller, higher-quality link program (focused on branded anchors and editorial placements in high-trust publications) alongside a much larger program of community mentions, creator placements, and review generation.
A useful rule: if a backlink campaign also produces a mention of your brand name inside the body copy of a credible publication, keep it. If it produces only an anchor-text link on a low-quality SEO network, the 2026 math says cut it and redirect that spend into mention-producing channels.
So what for your brand: Do not zero out your link budget. Reweight it. A 70/30 backlinks-to-mentions split from 2022 should look closer to 30/70 in 2026 for most mid-market brands.
Which channels actually produce brand mentions at scale
Answer capsule: Four channels carry almost all of the 2026 brand-mention gains: community conversations (Reddit, Quora, niche forums), creator and YouTube content, third-party reviews (G2, Capterra, Trustpilot), and earned editorial PR. Owned channels (your blog, newsletter, homepage) produce almost no AI citation lift because LLMs discount self-reported brand content relative to third-party mentions.
Platform citation data supports the ranking. Profound's AI platform analysis found Reddit is the leading source for both Google AI Overviews and Perplexity, and Wikipedia is the leading source for ChatGPT. Semrush's 100-million-citation study found Reddit, LinkedIn, YouTube, and Wikipedia together account for a disproportionate share of citations across ChatGPT, AI Overviews, and Perplexity. Reddit's own Ripple Effect research reports 88% of Reddit users say Reddit posts are "honest and truthful" - a trust premium that plausibly explains why AI models weight the source heavily.
What this means in practice:
- Community marketing is now the clearest mention-producing activity. A well-run Reddit or Quora program creates 30 to 150 named-mention moments per quarter in trusted threads, many of which Google indexes and AI models ingest.
- Creator and podcast PR converts to YouTube transcripts - the single highest-correlation signal in the Ahrefs data.
- Review presence on G2, Capterra, Trustpilot, and category-specific review sites gives ChatGPT and Perplexity a structured source for category queries.
- Earned editorial PR produces branded anchor mentions in high-trust domains. Wire distribution alone does not.
How much does a brand mention program cost?
Answer capsule: Mid-market community-and-mention programs run $3,000–$15,000 per month in retainer for one to three platforms, with a typical 6-month minimum engagement to hit the compounding window. Paid creator and PR layers run separately from retainer and typically range from $2,000–$20,000 per month depending on scope. The right benchmark is not CPM or CPC - it is mentions per quarter, categorized by source trust tier.
Pricing is driven by four variables: (1) number of target communities and platforms, (2) volume of content and threads per month, (3) whether reputation management is part of scope (active suppression of negative threads adds ~30% to cost), and (4) whether AI citation measurement tooling is included. For reference, an AI-visibility-only retainer (audit, prompt tracking, and content optimization) typically runs lower than a community-marketing retainer because it is less production-heavy; a combined program runs to the top of the range. Our own published guide on community marketing ROI walks through the 18-month value curve in CFO-legible terms.
For context on the CTR pressure that is forcing this budget re-weight: Seer Interactive's September 2025 analysis across 3,119 informational queries found organic CTR on AI Overview queries fell from 1.76% to 0.61% - a 61% drop. Paid CTR fell 68%. The money that used to buy clicks from ranking pages is increasingly buying nothing. Mention-producing activity is where the marginal dollar now earns its return.
So what for your brand: Budget a 6-month floor at $3K–$15K/month for community mentions and set a separate line for PR and creator work. Do not benchmark it against paid CPA. Benchmark it against the SEO ROI curve, where payback lands between months 6 and 12.
Who this budget shift is for - and who should wait
Answer capsule: Mid-market brands ($5M–$50M ARR) in consideration-heavy categories (SaaS, DTC, fintech, health, B2B services) see the highest return from shifting off-page spend toward mentions because their buyers validate decisions through community and AI. Enterprise brands and very early-stage brands need different allocations.
Conditional recommendations:
- Growth-stage SaaS and DTC ($5M–$50M ARR, selling into AI-mediated discovery): Reweight off-page budget toward 30/70 backlinks/mentions. Highest likely return on this shift.
- Enterprise brands (already high branded search volume, already cited in AI): The marginal mention is worth less because the model already knows the brand. Invest in defensive reputation management and category-mention depth rather than raw mention volume.
- Pre-$5M startups with limited budget: Do not run a full mention program yet. Focus on one community surface (usually Reddit or a niche forum) plus review placement. A light $2K–$4K/month engagement or an in-house operator is the right fit.
- Regulated categories (healthcare, fintech with compliance constraints): Mention production is slower and more expensive per unit. Allocate 12-month budgets, not quarterly ones, and tighten legal review into the workflow from day one.
So what for your brand: Match your allocation to your size and the mediation of your buying journey. Do not copy an enterprise SaaS off-page plan into a Series A budget.
How to measure a mention-led strategy
Answer capsule: Measure at four layers: (1) mention volume by source tier, (2) AI citation share on target prompts, (3) branded search volume growth, (4) revenue-linked outcomes (pipeline, CVR lift on branded queries, CAC delta). Track quarterly, not weekly. AI model retraining cycles move on 4–6 month windows.
A working reporting stack for a mid-market brand:
- Mention volume and source tier: Track with Ahrefs Brand Radar, Mention, or Brand24. Categorize by tier (Tier 1: named publications and Reddit/Quora on high-trust subreddits; Tier 2: mid-trust forums and niche blogs; Tier 3: low-quality syndication). Tier-1 mentions correlate with AI citations; Tier-3 do not.
- AI citation share: Track with Profound, Parse, Peec AI, or Evertune on the 20–50 prompts that matter for your category. The Wellows analysis emphasizes watching both explicit (with link) and implicit (without link) mentions to see the full signal.
- Branded search volume: Pull monthly from Google Search Console and Google Trends. Treat a 10%+ quarterly rise as a leading indicator of AI citation lift 2–3 quarters out. Our guide to auditing AI search visibility covers the prompt-building step in detail.
- Revenue linkage: Add a self-reported attribution question ("where did you hear about us?") and track CVR on branded search. In our client base, 18–35% of inbound demo requests cite a Reddit thread or AI answer when asked, even when those never appear in last-touch.
For a deeper build-out of the attribution side, our community marketing ROI model documents the multi-touch framework CFOs accept.
The 2026 off-page budget: a reweighting template
Answer capsule: For a typical mid-market SaaS or DTC brand spending $10K–$30K/month on off-page, the 2026 reweighting looks like this: cut generic link-building by half, route 40–60% of the reclaimed spend into community marketing (Reddit + Quora), and route the remainder into creator/YouTube PR and review generation. Keep a smaller, higher-quality link program focused on editorial placements with branded anchors.
| Budget line | 2022 typical share | 2026 recommended share | Rationale |
|---|---|---|---|
| Link-building (generic) | 50–70% | 15–25% | Low marginal AI visibility; retain for baseline ranking signal |
| Editorial PR / digital PR | 10–20% | 20–30% | Produces branded anchors; high trust editorial mentions |
| Community marketing | 0–5% | 25–40% | Highest mention volume per dollar; Reddit/Quora are top AI citation sources |
| Creator / YouTube PR | 0–10% | 15–20% | YouTube mentions correlate at ~0.737 - the single strongest signal |
| Review generation (G2/etc.) | 5–10% | 10–15% | Structured third-party source for category queries in ChatGPT and Perplexity |
The table is not a prescription. It is a starting point for the reweighting conversation, and it is the conversation your CFO is going to want to have once the CTR drop on your SEO-tracked queries hits the board deck.
FAQ
Are backlinks obsolete for AI visibility?
No. Backlinks still correlate positively with AI visibility (~0.218 per the Ahrefs 75K-brand study) and still move organic rankings, which is itself an upstream predictor of AI citations. The 2026 shift is that mentions have a 3x stronger correlation, so the marginal dollar buys more AI visibility in mention-producing channels.
What counts as a "brand mention" for AI visibility?
Any named reference to the brand in credible third-party content. This includes Reddit comments, Quora answers, YouTube video descriptions and transcripts, podcast show notes, news articles, Wikipedia entries, review sites (G2, Capterra, Trustpilot), and forum posts. Self-authored content on the brand's own domain does not count for this purpose.
How long does it take for new brand mentions to move AI citations?
Typically 3–6 months before model updates pick up new community content, and 4–12 months before the shift is visible across ChatGPT, Perplexity, and AI Overviews. Content updated within 30 days receives 3.2x more AI citations, but the lag on brand-level signal is longer because models retrain in cycles, not continuously.
Can I buy brand mentions the way I buy backlinks?
Some, but carefully. Paid creator and podcast placements produce mentions that correlate with AI visibility. Pay-for-play link networks and PR syndication without editorial pickup do not. Press releases on their own are cited in AI answers at roughly 0.04% - effectively zero. The filter is whether the placement includes a named, contextual mention from a source that LLMs treat as credible.
How does this change our SEO agency relationship?
Most traditional SEO agencies still price and measure off-page work on link count and domain authority. Ask them to add three things to your reporting: mentions by source tier, AI citation share on 20–50 target prompts, and branded search volume growth. If they cannot report on those, the reweighted budget should move to partners who can.
Does this apply to B2B as well as DTC?
Yes, with emphasis on different channels. B2B brands get disproportionate mention lift from LinkedIn, Quora, G2/Capterra reviews, and podcast appearances. DTC brands get more from Reddit, YouTube reviews, and Trustpilot. Both categories see the same directional budget shift away from generic link-building and toward earned-mention activity.
If your 2026 plan still leans on a link-building retainer but your category is being rewritten inside AI answers every quarter, the math is not going to improve on its own. We run the community and AI visibility layer for 280+ brands. Get an AI visibility audit and we will map your current mention surface against the prompts that matter for your category - then tell you, honestly, whether a budget reweight or a full program is the right move for your situation.