Brand Mentions vs Backlinks: What AI Search Rewards

AI engines answer from what independent pages say about you, so a mention on a page they can read does the job a backlink does for Google. In Ahrefs' study of 75,000 brands, branded web mentions correlated with visibility in Google AI Overviews at 0.664; backlinks scored 0.218. That is correlation, not proof, but the two studies we cite below point the same way.

Our position: stop treating mentions and links as rivals. An unlinked mention on a page an engine actually reads beats a link from a page it never retrieves. The work is getting independent, readable pages to name you next to your category.

What counts as a brand mention?

A brand mention is any appearance of your brand name on a page you do not control, with or without a hyperlink. A linked mention passes a page's authority to your URL. An unlinked mention only states your name in context. Both are earned media: coverage you did not pay for.

Mentions on your own site do not count as corroboration. They are your claim; the engines look for other people repeating it. Typical sources:

• Press coverage and "best of" roundups
• Review sites and marketplaces
• Comparison posts ("X vs Y")
• Forum threads, Reddit, Slack and Discord communities
• Podcast show notes, video descriptions and creator posts

Most of the last three arrive without a link. A Reddit reply naming your product rarely links to it, and a podcast host recommending your tool drops no URL.

Do brand mentions help your Google rankings?

Only modestly, and nobody has shown more. Google's documentation says it uses links as a signal when determining the relevancy of pages; it does not describe unlinked mentions as a ranking signal. For competitive Google queries, keep building links.

Our reading is that consistent mentions help Google connect your name to a topic. That is plausible, undocumented and small. The two also overlap: in the same Ahrefs data, brand-name anchor text correlated at 0.527, and a link that names you is a mention.

Why do mentions matter more in AI search?

Generative engines synthesize an answer from many sources instead of ranking one page, so whichever brands those sources keep naming get repeated. Two studies show how lopsided that gets.

• Source type. In a September 2025 arXiv study, AI search leaned hard toward earned media. For consumer electronics queries in Canada, 77.6% of an AI search engine's sources were earned and 0.3% social. Google's mix was 54.1% earned and 23.1% social. The authors used ranking-style prompts and one search-enabled model, gpt-4o-search-preview, so treat it as a strong signal, not a universal law.
• Volume. Ahrefs sorted brands by web mentions. The top quartile averaged 169 AI Overview mentions, against 14 for the next quartile and 0–3 for the bottom two. Twenty-six percent of brands had no AI Overview mentions at all.

The pattern looks like a threshold, not a slope: below a certain amount of independent talk, a brand barely registers. Ahrefs itself warns that correlation is not causation. For the mechanism, see how AI assistants decide which brands to recommend.

Do backlinks still matter for AI answers?

Yes for Google's AI Overviews, far less for assistants such as ChatGPT and Gemini. Ahrefs found that 76% of AI Overview citations come from top-10 pages, but only about 12% of URLs cited by the assistants it tested also ranked in Google's top 10. Perplexity was highest at 28.6%; the rest hovered near 8%.

So split the goal by engine. To appear in AI Overviews, a page must be indexed and eligible to show a snippet, so classic ranking work still pays. For the other assistants, rankings predict little, and independent mentions are the lever you can pull.

Which mentions can AI engines actually read?

Only mentions on pages the engine can fetch or has already learned from. That is less certain than it sounds, because the sources AI leans on are the ones most likely to restrict AI crawlers.

In AEOeye's crawler access study (October 2026, 384 sites with readable or absent robots.txt), 27 of 38 news publishers (71%) blocked at least one AI retrieval bot, against 1 of 193 SaaS brands. Five of 20 review and community sites did too. Reddit's robots.txt is two lines, User-agent: * and Disallow: /, with access governed by its Public Content Policy and licensing deals. The G2 and Capterra homepages returned 403 to our self-identified research crawler.

Limits: a convenience sample, one fetch per site, and robots.txt is a request, not enforcement. The raw data is in the per-site CSV.

OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. So before paying for a placement, check the host's robots.txt for OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot.

How do you earn mentions AI engines read?

Start from the pages AI already cites for your category, then earn a place on them. A press hit on a host the engines cannot read is a worse bet than a modest comparison post they already cite.

1. Map the sources. Ask each engine your buyers' questions and list every page it cites for competitors. That list is your target list.
2. Be easy to describe correctly. Comparison and review writers copy what is easy to find: a public pricing page, a plain feature list, a spec table.
3. Publish something quotable. In the GEO paper, the top methods, adding statistics, quotations and citations to content, raised visibility by 30–40% on its position-adjusted word-count metric. Original data gives writers something to cite; our crawler study ships with its raw CSV so others can check and quote it.
4. Pitch writers, not link farms. Target journalists and newsletter authors who cover your category. A founder quote is a mention whether or not it is linked.
5. Answer in communities under your own name. Be useful first and let the mention follow. We cover the mechanics in Reddit SEO.
6. Keep one spelling everywhere. Use identical capitalization across bios and listings, and add sameAs links to your official profiles in Organization markup; schema.org defines sameAs as a page that unambiguously indicates an item's identity. In our study, 190 of the 312 homepages we could fetch carried them.

How do you track brand mentions?

Track two layers separately: mentions on the open web, which listening tools can monitor, and mentions inside AI answers, which only a direct question reveals. Ahrefs' study describes Brand Radar's web-visibility view for the first layer. No dashboard of web mentions tells you what ChatGPT said yesterday.

For the second layer, run this protocol:

1. Write 10 unbranded buyer questions, such as "best invoicing tool for freelancers".
2. Ask each in every engine you care about.
3. Repeat each question several times. SparkToro had 600 volunteers run 12 prompts 2,961 times and found under a 1-in-100 chance that ChatGPT or Google's AI returns the same brand list twice. It also found that visibility percentage across many prompts is a decent proxy.
4. Log whether you are named, in what position, which competitors appear and which sources are cited.
5. Feed the cited sources back into step 1 of earning.

AEOeye's free audit does a version of this: three unbranded buyer questions asked to ChatGPT live, the real answers shown in full, and the competitors named instead. The $29 full report adds Claude, Gemini, Google AI and Perplexity, and ten questions. For daily monitoring across hundreds of prompts, dedicated trackers fit better; see our tracker comparison.

If the audit comes back quiet, why AI does not mention your brand is the next read.