LLMention
← Back to ScannerAbout

About LLMention

A free scanner that tells you whether AI search engines can read, parse and cite your pages, and publishes the rules it uses to decide.

Browse sections

Why this exists

People increasingly get answers from an assistant instead of clicking a link. When that happens, a page that ranks well but is never quoted has lost the visitor entirely, and the usual analytics will not tell you why. LLMention exists to make that failure mode visible before it costs you traffic.

The category has a credibility problem in the other direction too. Tools describe mysterious “AI visibility scores” without saying how they are computed, and some make claims about files and markup that the evidence does not support. This project takes the opposite approach.

What makes it different

  • The method is published in full. All twelve dimensions, every check, its exact rule and its point value are on the methodology page, generated from the same data the scanner executes, so the documentation cannot drift from the behaviour.
  • No score floors. A page that satisfies nothing scores near zero rather than being lifted to a respectable-looking minimum.
  • Low-value signals are weighted low. llms.txt is one of the three least-weighted dimensions here: the check is worth 4 of that dimension’s 6 points and the dimension carries 5% of the total, so the file is capped at 3.3 of the 100.
  • Limits are stated up front. The scanner does not query ChatGPT about your brand. It measures whether your pages are in a state that makes being cited possible, which is a narrower and more honest claim.

Who runs it

LLMention is maintained by Alex Xu as an independent project. He is an independent developer based in China; this is not a company, and the site is not affiliated with OpenAI, Anthropic, Google or Perplexity or holding any data relationship with them. The scanner is free and requires no account.

It is not a free site with nothing behind it, so the funding is stated rather than left to be discovered: advertising served by Google, and a paid measurement report described on the pricing page. Neither can change a score — the score is computed by the scanner from the response it fetches, and a report explains a result rather than editing one. The author page sets out the editorial rules and the conflicts that come with that model.

Written and maintained by Alex Xu. The scoring method is published in full on the methodology page, the source is public at github.com/Alex13192/geo-scanner, so a disagreement with a result can be specific: say which check and which URL. The longer-form writing — what the study found, and where the method is weaker than it looks — is at /blog/.

The name, and what it is not

The product’s full name is LLMention GEO Scanner. The free scanner is an independent readiness audit — it fetches your pages the way a crawler would and reports whether they can be read and cited, and it does not ask an engine about your brand. The paid audit measures what the engines actually say: 25–30 buyer questions, three runs each, the answers recorded and the mentions counted. Neither product measures a ranking or predicts traffic.

It is not affiliated with DataForSEO’s LLM Mentions API — the product at llm.dataforseo.com — which measures a different thing: there is no partnership, data relationship, or shared operator between the two.

Contact

Questions and corrections both go to the same place: services@llmention-geo.com. If the scanner reports something about your site that you believe is wrong, that is the most useful message you can send, and the method is public specifically so a disagreement can be specific.

Primary sources

Where this project makes a judgement about what generative engines favour, it follows published research rather than folklore:

What the scanner does and does not do

In the published study over 30 homepages, 2 of them (7%) reached a B or above and none reached an A. That result is easier to read with the scope stated plainly.

It doesIt does not
Fetch your homepage the way a crawler wouldExecute JavaScript, so a client-rendered page is judged as a crawler sees it
Read robots.txt, llms.txt and the sitemap when they existSign in, submit a form, or reach anything behind a paywall
Report each failed check with the evidence that produced itAsk any AI engine whether your brand is mentioned, recommended or cited
Publish every rule, its weight and its pass conditionPredict a citation, or promise a ranking
Two colleagues in casual clothes talking over a laptop at a wooden table in a plant-filled officeA team member presenting charts on a large screen to seated colleagues in a bright meeting room
Illustrative photography (Unsplash License) — not a photograph of the maintainer, and no client or employee is depicted or implied. Credits: public/photos/CREDITS.md.

Questions about the project

Who runs LLMention?

Alex, an independent developer, named on the author page with the GitHub account the project is published under. It is not affiliated with OpenAI, Anthropic, Google or Perplexity, and holds no data relationship with any of them.

Is LLMention the same as DataForSEO's LLM Mentions API?

No. The full name is LLMention GEO Scanner, an independent readiness audit; it does not track mentions or query an engine about a brand. DataForSEO's LLM Mentions API is a different product that measures mentions. There is no partnership, data relationship or shared operator between the two.

Why publish the scoring method in full?

Because the category has a credibility problem in the other direction: tools describe mysterious AI visibility scores without saying how they are computed. A method you cannot inspect is not evidence.

Does the scanner query ChatGPT about my brand?

No. It measures whether your pages are in a state that makes being cited possible, which is a narrower claim than whether you are being cited, and a more defensible one.

Evidence and sources

GEO can boost visibility by up to 40% overall in generative engine responses, and in the paper's Table 1 the best of the tested methods improve on the no-optimization baseline by 41% and 28%. — Generative Engine Optimization, KDD 2024

The weightings on this site follow that measurement rather than taste, and the parts of the picture a single-URL scan cannot see are stated rather than left out.

Primary sources