Brand prominence: definition, citation concentration and the confounder

Brand prominence is the degree to which a brand or publisher is already known, mentioned and cited across the web. It is a property of the brand, not of any single page. You can't read it from a page crawl, and no engine publishes a number for it.

In AI visibility work the term covers two related ideas:

• Brand size effect: large, established brands appear in AI answers far more often than small ones.
• Citation concentration: a few sources capture most of the citations an engine hands out.

The clearest measurement comes from a study of news sources in AI search (arXiv:2507.05301). It found that the top 20 most frequent news sources accounted for 67.3% of all citations for OpenAI models. The same study reported a Gini coefficient of 0.83 for OpenAI, 0.77 for Perplexity and 0.69 for Google, where higher means more concentrated. Our digest of the news citing patterns study covers the method.

Two caveats matter. The data window was March to May 2025, so it is old as of October 2026. It also came from AI Search Arena, a self-selected evaluation population, not ordinary user traffic.

The pattern is consistent elsewhere. The 252,000-trial factorial study (arXiv:2605.25517) anonymised brands by design and warned that real systems may favour established domains. For the plain-English version, see whether AI chatbots favor big brands.

Prominence is the dominant uncontrolled confounder in correlational GEO studies. A study notices that cited pages share some trait, then credits the trait. But large brands have more reviews, more backlinks, more press and more pages. Any of those traits will correlate with citations because prominence drives both.

The usual fix fails. Researchers often control for Ahrefs DR or Moz DA. Those are vendor regressions built from backlinks and traffic, which are proxies for prominence itself. Conditioning on a mis-specified proxy doesn't remove the confound.

What this means for you:

1. Treat any "brands with X get cited more" claim as probably measuring size.
2. Benchmark yourself against brands of similar size, not category leaders.
3. Work on what you control. Access levers (server-side rendering, per-bot robots.txt, firewall reachability, snippet controls) decide whether your pages can be used at all.
4. Where engines cite other sites instead of yours, being named on those sites is associated with appearing in the answer. It is an association, not a promised effect. See which sites AI cites most.

"Build domain authority to get cited"

Domain Authority is a Moz metric. No AI engine computes it, and Google publishes no site-level authority score. There is nothing to build toward.

"Prominence is a ranking factor you can optimize"

It is a strong association with no documented lever. You can't move it this quarter through markup or phrasing.

"Training crawlers are how prominence gets into the model"

Training inclusion is a theoretical path into parametric answers, but its effect is unmeasured. Read whether blocking training bots affects brand recall before changing robots.txt.

"One visibility score shows how I compare"

A blended score across platforms, read against bigger brands, mostly reports brand size wearing a GEO label.

• Position bias: the strong effect of where a document sits in the model's context, which also has no publisher lever.
• Predictable prose and AI citation: another observational finding confounded with editorial budget and brand size.
• Topical relevance: the one content factor that several studies, including the critical survey of GEO research (arXiv:2607.14035), agree is reproducible.

What is citation concentration? Citation concentration describes how unevenly an AI engine spreads its citations across sources. When a few sources take most citations, concentration is high. The news citing study (arXiv:2507.05301) measured this with a Gini coefficient and found OpenAI models at 0.83, Perplexity at 0.77 and Google at 0.69, using data from March to May 2025.

Can a small brand overcome the brand size effect? Nobody has measured a reliable way to do it, so no one can promise it. What you control is access: make sure AI crawlers can reach your pages, read your key facts without JavaScript, and that snippet controls aren't hiding content. Being named on sites that engines already cite is associated with appearing in answers, without a guaranteed effect.

Why can't researchers just control for brand size? Because the available controls are vendor metrics like Domain Rating and Domain Authority, built from backlinks and traffic. Those are proxies for prominence itself. Adjusting for a flawed proxy leaves most of the confound in place. The honest approach is to stratify, comparing like-sized brands, and to report results within each size band.

How should I benchmark my own AI visibility? Compare yourself with competitors of similar size, measure each platform separately, and ask the questions your buyers actually ask. The free checker on this site gives a snapshot for up to three buyer questions. Read any result as a sample of answers, not as a rank or a measure of your brand's worth.