You are hiring, and two candidates hand you a CV. Both CVs claim excellence. One candidate also has six people in your industry who, unprompted, describe their work in similar terms.
You do not believe the CV. You believe the corroboration. Not because the CV is lying, but because a claim that only exists in one place — the place with an interest in making it — is not evidence of anything.
An answer engine assembles a picture of you from many sources at once. Your website is the CV. Everything else is the reference check. Most businesses spend ninety per cent of their marketing effort polishing the CV.
The evidence
Ahrefs studied 75,000 brands, measuring which signals correlate with being mentioned in Google's AI Overviews. They used Spearman rank correlation and published the full table. This is the single most useful dataset in the field. TIER A
| Signal | Correlation | Read it as |
|---|---|---|
| Branded web mentions | 0.664 | How often your name appears anywhere online |
| Branded anchor text | 0.527 | Links that use your actual name as the link text |
| Branded search volume | 0.392 | How many people search for you by name |
| Domain Rating | 0.326 | Classic authority score |
| Referring domains | 0.295 | How many sites link to you |
| Branded traffic | 0.274 | Visits from people who searched your name |
| Total backlinks | 0.218 | The thing SEO spent twenty years on |
| URL rating | 0.180 | Page-level authority |
| Number of pages on site | 0.170 | Publishing volume |
Ahrefs, "An Analysis of AI Overview Brand Visibility Factors", 75,000 brands, Spearman correlation. A companion Ahrefs report found YouTube mentions — brand name appearing in a video title, description or transcript — correlating at approximately 0.737, the highest single figure they published. TIER B
The headline: brand mentions correlate roughly three times more strongly with AI visibility than backlinks do. For twenty years, an unlinked mention of your company was considered worthless by SEO orthodoxy — a wasted opportunity, something to chase down and convert into a link. That orthodoxy is now backwards.
Ahrefs themselves say it: correlation is not causation, and they describe these as moderate-to-weak relationships. Big, well-known brands get mentioned a lot and appear in AI answers a lot, and both may be caused by simply being a big, well-known brand. Do not treat 0.664 as a dial you can turn.
What makes the finding actionable anyway is that it agrees with the mechanism from Chapter 1. Retrieval systems assemble answers from multiple sources and reward agreement between them. A model with no corroborating evidence about you has nothing to be confident about. The correlation and the mechanism point the same way, which is the strongest form of evidence available in a field this young.
What to do about it
The instruction "get mentioned more" is useless without a method. Here is the method, in rough order of return for a small or mid-sized business.
Industry associations, chambers, trade bodies, licensing registers, local business directories, LinkedIn company page, Crunchbase, Google Business Profile, Apple Business Connect. These are boring, free, permanent, and they are the cheapest corroboration you will ever buy. Do all of them before you do anything clever.
For software: G2, Capterra, TrustRadius. For services: Google reviews, industry-specific directories. Review platforms appeared in roughly 34.5% of commercial AI Overviews in one vendor analysis TIER B — and structurally it makes sense, because they are the only third-party source with organised, current, category-level data on vendors.
Journalist-request platforms, trade press, podcast guesting, industry panels, association newsletters. The objective is a named mention in a piece of content someone else published. You do not need a link. You need your name, in context, next to your category.
Given the strength of the YouTube signal, and given that AI engines can read titles, descriptions and transcripts, a modest channel of genuinely useful videos is disproportionately valuable. This is not about production quality or subscriber counts. It is about your brand name existing inside a transcript on a platform the engines trust.
Reddit, specialist forums, LinkedIn. Chapter 12 covers this properly, including how to do it without getting banned and without embarrassing yourself.
Google's own guidance names this explicitly as a myth to avoid: seeking inauthentic mentions across the web does not work, and their systems are built to filter manipulation. TIER A Paid mention networks, spun press releases pushed to two hundred syndication sites, and AI-generated "citations" on sites nobody reads are the 2026 equivalent of buying links in 2011. They failed then. They are failing faster now, because agreement between low-quality sources is exactly the pattern these systems are tuned to discount.
- Search Google for your brand name in quotes, minus your own domain:
"Your Company" -site:yourdomain.com. Count the genuinely independent results on the first three pages. That number is your starting position. - Do the same for your two strongest competitors. The gap is your work list.
- Ask an AI directly: "What do you know about [company name]?" The length, accuracy and confidence of the answer is a rough proxy for how much corroboration exists.
Spend one working day claiming and fully completing every free profile, directory and register that applies to your business, with identical wording for your name, address and description on each. It is dull, it is free, and it is the highest-return day in this book.