If I'm being honest, I used to file digital PR under "fluff." Placements you can't attribute, mentions with no link, a pretty coverage report nobody opens. Performance marketing brain. Then I watched a brand with mediocre rankings and a thin backlink profile get named in ChatGPT answer after ChatGPT answer, and I went looking for why. The why turned out to be twelve unlinked mentions in trade publications.
That experience has data behind it now, and the data is lopsided. Ahrefs studied 75,000 brands and found web mentions correlate with AI visibility at 0.664. Backlinks? 0.218. The signal SEO spent twenty years optimizing is three times weaker, in AI answers, than the signal PR was generating all along. SEO is turning into PR, and most SaaS teams have the budget pointed at the wrong one.
Here's the playbook for pointing it at the right one.
Key Takeaways
- Ahrefs' 75,000-brand study found mentions correlate with AI Overview visibility at 0.664 versus 0.218 for backlinks. Language models learn brands from words, with or without links.
- Nofollow and dofollow links show nearly identical AI influence, so link-equity tactics transfer almost nothing to citation visibility.
- About 85% of brand mentions in AI answers come from third-party sources, and roughly 90% of those live in listicles, comparisons, and reviews, per AirOps' 21,311-mention study.
- Mention quality is pitchable: brand plus category in the same sentence, a stat attached to your name, and consistent facts across placements.
- Google reports none of this, so mention-to-citation tracking is the only way to prove PR's AI value.
The Evidence: AI Answers Run on Mentions
The claim sounds like PR-agency wishful thinking, so let's stack the receipts.
Language models build brand knowledge from text they've read: articles, roundups, forum posts, transcripts. A link is a navigation instruction; a mention is information. Ahrefs' correlation study made that concrete at scale, and it came with a detail that should genuinely rattle link builders: nofollow and dofollow links showed nearly identical influence on AI visibility. The distinction Google taught us to obsess over is invisible to the machines now answering buyer questions. Even Rand Fishkin, who helped build the link-obsessed era, has called the turn:
"Godfather of SEO, @randfish: “Mentions matter more than backlinks. If you’re still link building, I have great news, you can stop. You don’t have to do it. It’s like washing chicken, you don’t have to do it man, just sto..." — @edwardeachday, July 2026
His argument, and I'll paraphrase so this survives the internet: mentions matter more than backlinks now, and the practice of chasing links for their own sake is winding down. Meanwhile Search Engine Land's work on unlinked mentions notes that nearly nine in ten pages ChatGPT cites sit outside the top 20 organic results. The AI answer layer isn't reading your rankings. It's reading everything else.
And the surface this feeds is growing fast: AI search visits jumped 42.8% year over year, from 15.6 billion to 27.4 billion quarterly, per Wix's AI Search Lab data. The mention economy isn't a side channel anymore. It's where a growing share of your category's questions get answered.
One stat from the Ahrefs data doubles as a warning: 26% of brands studied had zero AI Overview presence at all. Not low. Zero. Those are companies with websites, rankings, and link profiles, and no textual footprint the models found worth repeating. We wrote about the backlink side of this in our own 20-site backlink test; the mention side is the half most teams still haven't priced in.
AI search traffic grew 835% this year. Is your content ready?
RankControl generates 15+ content types optimized for ChatGPT, Claude, and Perplexity. Published on your domain, matched to your brand.
Start publishing→
Where Mentions Convert: Not All Coverage Is Equal
Funny enough, the best study on this question wasn't run by a PR firm. AirOps analyzed 21,311 brand mentions across ChatGPT, Claude, and Perplexity and found 85% came from third-party sources, with brands 6.5x more likely to be mentioned via someone else's page than their own.
The breakdown of that 85% is the tactical gold: roughly 90% of third-party mentions live in listicles, comparisons, and reviews. Not feature profiles. Not funding announcements. The unglamorous "best X for Y" roundup is the single highest-converting placement type in AI search, which is exactly what our listicle strategy guide predicted from the content side. If a PR campaign gets you one profile in a business magazine or three slots in mid-tier category roundups, take the roundups every time.
Two more findings shape the targeting:
- 68% of brand mentions were unique to a single AI model. Visibility on Perplexity says nothing about ChatGPT. Spreading placements across publication types hedges the platform lottery.
- Presence on 4+ platforms correlates with 2.8x higher odds of appearing in ChatGPT responses, per Wix's AI Search Lab data covered by Contently. Podcasts with transcripts, newsletters, trade blogs, and communities all count as platforms. Breadth beats one big hit.
And it does work when pointed at AEO deliberately: one agency's documented campaign landed 151 placements across 92 outlets and watched daily Bing AI citations roughly double. Single case, self-reported, but directionally consistent with everything above.
The New PR Brief: Pitch for Mention Quality
Actually, rewind, because the biggest change isn't where you pitch. It's what you ask for. The old brief optimized for a followed link and any old anchor text. The new brief optimizes for how the sentence containing your brand reads, because that sentence is what a model ingests. Four asks, in priority order:
- Brand and category in the same breath. "RankControl, an AI search visibility platform," beats a naked name-drop. Co-occurrence of brand and category terms is how models learn what you are; Seer Interactive's fan-out analysis shows brands that co-occur with category language surface across the sub-queries AI engines actually run.
- A stat glued to your name. "Customers using X cut reporting time 40%" is a liftable fact with your brand inside it. Numbers travel; adjectives don't.
- Consistent facts everywhere. Same category label, same claim, same numbers across every placement. Models trust claims that repeat consistently across independent sources, and contradictions read as noise. One canonical boilerplate, enforced ruthlessly (this is precisely what our Brand Control profile keeps synced for customers, because drift creeps in one podcast intro at a time).
- Tables when you can get them. Comparison tables get extracted at 81% versus 23% for prose in Contently's reporting. A row in a comparison table is a structured fact about you on someone else's trusted domain.
Notice what's not on the list: a link. Take links when offered, obviously. Just stop trading away mention quality to get them.
The reactive channels still work for volume: the HARO successors (Qwoted, Connectively, Featured) put you in front of journalists writing the listicles and roundups models feed on. Expect single-digit win rates and treat every win as a compounding textual asset.
Built by the team that got cited in 48 hours.
Content generation, backlink building, AI visibility tracking, and lead capture. One platform, zero guesswork.
See the platform→Mine the Mentions You Already Have
Before buying new coverage, look, you're probably sitting on an unworked pile of it. Every brand that's been around a few years has unlinked mentions scattered across the web, and the classic playbook treated them as link-reclamation targets: email the author, beg for the link. The AEO playbook is different, and honestly easier, because the mention already did the main job.
The workflow: run your brand name through a content index (Ahrefs Content Explorer with your own domain excluded is the standard move), filter for pages with real authority and organic traffic, and audit what those sentences actually say about you. Then upgrade, in this order: fix wrong facts first (a stale "starting at $99/mo" in a well-ranked roundup will get repeated by a model verbatim), request category framing second, and ask for the link last, as a bonus rather than the point. Set alerts so new mentions get the same treatment while they're fresh, and re-audit quarterly; alerts alone miss about a third of what a periodic sweep catches.
We've seen the payoff pattern repeatedly: a brand gets named in a handful of "best tools" threads and city-level listicles, no links anywhere, and starts appearing in AI answers for exactly those categories within weeks. The client's first question is always "but where's the link?" The mention was the asset. There was never going to be a link, and it never mattered.
Who Owns This? (The Gap Where the Value Leaks)
The organizational problem might be bigger than the tactical one. Earned mentions drive AI citations, but mentions don't fall under SEO, and AI citations don't fall under PR. This thread captures teams hitting that wall in real time:
The consensus in there matches what we hear from customers: PR lands the coverage, SEO reports that rankings didn't move, and the AI citations the campaign actually produced show up in nobody's dashboard. Google has started reporting impressions for linked URLs in AI features, but unlinked mentions, the most valuable currency in this whole system, appear nowhere. Practitioners running GEO programs have been loudly pointing at the same hole: the data teams need to prove PR's AI value simply isn't in any Google product.
So the metric has to be built independently: mention-to-citation conversion. Log placements as they land, with date, outlet, category framing, and the exact sentence your brand sits in. Sample your category queries across ChatGPT, Perplexity, Gemini, and Copilot every week. When your brand enters answers for the queries a placement targeted, that's the conversion, and it's attributable in a way no coverage report ever was. This is exactly the loop our AI visibility tracking automates: every tracked query, every engine, sampled continuously, so a March placement that starts paying off in May gets credited instead of forgotten.
Worth adding up what the manual version costs, because the hours hide in the recurring parts. The initial unlinked-mention audit runs 4-6 hours for an established brand. Fact-correction outreach: 2-3 hours per batch. Weekly citation sampling across twenty queries and four engines: 3-4 hours, every week, indefinitely. Reactive pitching to keep new mentions flowing: another 2-3 hours weekly. Call it 6-8 hours a week of steady-state work after a 10-hour setup. Founders do the first month enthusiastically. Month three is where the program quietly dies.
The uncomfortable truth about mentions is that they decay silently too. Models retrain, roundups get rewritten, and last year's textual footprint fades. The real problem isn't earning the mention once. It's knowing when the machine stops repeating it.
See your first AI citation report in under 5 minutes.
No setup calls. No onboarding meetings. Connect your domain and see where AI mentions your brand right now.
Get started→
Point the Budget at Words, Not Links
The whole shift in one sentence: AI engines repeat what trusted text says about you, so the job is getting into trusted text, saying the right thing, everywhere, consistently. Start with the audit of mentions you already own, fix the facts, then pitch the roundups and comparisons with the new brief: category framing, a stat on your name, no begging for links.
You can run the audits, the pitches, the boilerplate policing, and the weekly citation sampling yourself. Or RankControl's agents can keep your brand facts consistent, track every mention, and show you which ones actually turned into AI citations, while you go be worth mentioning.
