GEO
Google's search ignores llms.txt. Independent data shows 97% of the files are never fetched. We still score it 5 points out of 100. Here is the working behind that number, and the three things that move AI visibility more.
Is llms.txt worth it? We get asked most weeks, usually by an owner who has just been quoted real money for an "AI visibility file". The honest answer is awkward. Google says its search ignores llms.txt. Independent log data shows 97% of llms.txt files are never fetched by crawlers. And we still award the file 5 points out of 100 in our readiness score. All three of those things are true at once. This post shows the working, because the working is more useful than the verdict.
llms.txt is a plain text file that lives at the root of your website, at an address like yourdomain.com/llms.txt. You write it yourself. It tells AI systems what your site is, what you offer, and which pages carry the substance, in a format machines can read without wading through menus, cookie banners and scripts. Think of it as a welcome note pinned by the front door for AI crawlers (a crawler is the software an AI company sends out to read web pages). The idea came out of the AI developer community in late 2024. No major engine has ever formally committed to reading it.
In June 2026, Google updated its Search Central documentation (the official manual for how Google treats websites) to state that Google Search does not use llms.txt, including its AI features. Keeping the file will not harm your visibility in Google. It will not help it either.
Google's John Mueller put it more bluntly, comparing llms.txt to the keywords meta tag: a self-reported label that search engines learned to ignore long ago, because anyone can claim anything about their own site.
Then the confusing part. Chrome's Lighthouse tool (Google's own site auditing software) moved an llms.txt check into its default audits in version 13.3. So one Google team now flags a file that another Google team says it ignores. If Google cannot agree with itself, a business owner can be forgiven for being unsure.
A file only matters if something reads it. The independent data says almost nothing does.
Digital Applied's llms.txt study tracked fetch behaviour and found that 97% of the llms.txt files it followed received zero requests from crawlers. Not low traffic. Zero. Nothing fetched them, human or bot.
Adoption tells the same story. The same study found that 8.7% of the Tranco top 1,000 published an llms.txt file as of June 2026 (Tranco is a research ranking of the most popular sites on the web). That is 87 sites out of the 1,000 with the most to gain. The rest have not bothered.
One caveat we should own: fetch data measures whether the file was requested, not whether it changed any answer. A fetched file might still do nothing. But the unfetched 97% cannot be doing anything at all.
Our readiness score asks one question: if an AI engine arrived at your site today, could it read you and understand you? Readiness is "is the door unlocked". Visibility, whether engines actually talk about you, is "is anyone walking through it". In that score, llms.txt earns a +5 bonus out of 100. Not zero, and not more than 5. Here is the reasoning at both edges.
Why not zero:
Why not more than 5:
The evidence supports a small bonus, so a small bonus is what we give. It follows that anyone selling an "llms.txt service" as a visibility fix is selling snake oil. Five points is not a fix. It is a rounding error with good manners.
We publish our own file at whocanfindme.com/llms.txt, because we practise what we score. The full weighting behind every point is on our methodology page.
If you have one afternoon for AI visibility, spend it on these three things, in this order. Each is worth far more than 5 points.
Nothing else matters if ChatGPT, Perplexity, Gemini and Claude cannot read your site. Your robots.txt file (the standard file that tells crawlers what they may and may not read) or your hosting firewall may be blocking them without you ever choosing to. We have seen sites blocking ChatGPT without knowing it. You can check yours in about a minute with our free crawler check.
Engines read your pages as text. If your content only appears after scripts run in a browser, some engines never see it at all. We wrote up what AI search engines actually read on your site, and the gap between what you see and what they see is usually the biggest surprise in a scan.
Engines learn who you are from pages they already trust: directories, review sites, local press, community threads. A consistent name, category and description across those places does more for your visibility than any self-reported file on your own domain ever will.
llms.txt is a 5-point bonus, priced by evidence. Publish one, take the points, and then put your real effort into access, extractable pages and mentions. If you want to know where you stand before you touch anything: run a free scan at whocanfindme.com. No signup. Takes about ten seconds.
When your brand name shows up in an AI answer, one of three things has happened: a real mention, an echo of your own question, or an answer about a different company with a similar name. Only the first is visibility. We classify every answer and score the other two as zero.
Leer másChatGPT mentions us in 58% of answers and recommends us in 0%. The gap between being seen and being chosen is where AI search is won, and almost nobody measures it.
Leer más