Brand Mentions and AI Visibility: Evidence and a PR Work Plan

CH.01

What the Ahrefs Correlations Measure

Ahrefs reports Spearman correlations of 0.664 for web mentions and 0.218 for backlink count with brand visibility in Google AI Overviews. These are associations, not citation probabilities, algorithm weights or campaign returns.

The May 26, 2025 study initially selected 75,000 brands, using domains with DR above 40 and a highest-volume keyword with at least 800 monthly searches. Around 26% had no AI Overview mentions and were excluded from the correlation analysis. That selection limits generalization to other brands, products or periods.

Web mentions include linked and unlinked mentions. This is not an isolated experiment on unlinked PR coverage. Dividing the coefficients does not show a threefold increase in recommendations or ROI, and it does not determine how your budget should move. Evaluate audience fit, cost and observed outcomes.

CH.02

Evaluate PR Without an Arbitrage Claim

A media strategy needs an audience, an editorial reason and a measurable objective. Different contribution and adoption denominators cannot establish an untapped return premium.

Relevant independent reporting can help people check a brand’s claims. That practical value does not reveal how a model scores trust, whether all mentions are independent, or why an answer selected a source. Paid or repeated coverage should not be presented as independent endorsement.

Choose a bounded PR pilot for a real audience and question. State the work, cost, evidence and review period before starting. Judge placement quality and observed outcomes; do not assume scarcity, a closing window or a threefold advantage from a correlation.

CH.03

Include Third-Party and Owned Sources

An answer may mention your brand using an external page, your own page or no visible source. Inspect the actual answer and source instead of assuming a universal third-party share.

Record which page is displayed as a citation, whether it mentions the right brand, and what it actually supports. A brand mention is not necessarily a link to your site, and a cited page is not proof of how the model learned the fact. Source mix varies with the prompt and product.

Keep owned information accurate and offer useful material to relevant publishers. Independent coverage may complement your site; it does not justify a universal citation lift or abandoning owned content. Verify each placement and observation within its conditions.

CH.04

Build an Editorial PR Workflow

Add story development and source verification to off-page work. The mix of link outreach, owned content and media relations should follow audience needs and measured business objectives.

Create something an editor can responsibly use: relevant facts, an accountable expert and a documented method. An earned article may contain a link or just a mention. Neither format automatically wins AI visibility, and the Ahrefs correlations do not measure a team reorganization.

CH.05

Connect Observations Without Claiming Causal Attribution

A visible source URL can be matched to a placement record. That is traceability, not proof that the PR effort caused the mention, recommendation or business result.

Save the prompt, product and mode, date, language and region, repeated observations and visible source URLs. An answer may omit sources or cite a page that does not support the brand claim. Record unknowns instead of assuming the model’s training data or a complete causal chain.

CH.06

The 90-Day Digital PR Launch: An Actionable Cadence for Earning Mentions

This 90-day, three-phase schedule is a Tenten planning option, not a deadline for coverage or AI results. Adapt it to evidence readiness, editorial availability and your team’s capacity.

Start with useful material and relevant audiences, then review what was actually published and observed. More platforms do not automatically mean greater recommendation probability; no fixed platform quota determines the plan.