Does Your Website Need llms.txt? What Server Logs Say

You've probably been told your website needs an llms.txt file to show up in ChatGPT. Some website builders now list it as a feature. We took it off our pricing page, and Tilazo sites don't ship one. This post explains why, using the only evidence that settles the question: server logs that show what AI crawlers actually request.

Short version: llms.txt is harmless, but AI crawlers almost never ask for it. Your time is better spent on the things they do read.

What llms.txt is

llms.txt is a proposed plain-text file that sits at yoursite.com/llms.txt. It's meant to give AI tools a tidy summary of your site and a list of your most important pages, a bit like a table of contents written for machines.

It's only a proposal. No AI company has said its search product reads it, and a file only helps if something asks for it. Every request a crawler makes shows up in your server logs, so you can check whether anything does.

What the server logs say

Three separate groups have now looked at this, at very different scales. They found the same thing.

Limy: 408 requests out of more than 500 million. Limy, an AI search analytics company, reports that it monitored over 500 million AI bot visits across a 90-day window, and only 408 targeted llms.txt directly. That's well under a thousandth of a percent. It's the company's own monitoring and it hasn't published a full method, so treat it as one data point, not the last word.

OtterlyAI: 84 visits on a site built to be found. OtterlyAI ran a 90-day server-log test on a site with a correctly implemented llms.txt file. Of more than 62,100 AI bot visits, 84 requested llms.txt, about 0.1% of AI bot traffic. Their conclusion was that the file had no significant impact on AI crawler behavior.

Ahrefs: 97% of files got zero requests. Ahrefs tracked 137,210 domains in its web analytics and found that 97% of llms.txt files received zero requests in May 2026. Of the requests that did happen, 19.5% came from AI bots, and the rest came from other bots and tools. GPTBot was the top AI bot and Claude Code was second. GPTBot collects content for training OpenAI's models. It isn't the crawler that decides whether ChatGPT search can show your site.

Study	What was measured	Requests for llms.txt
Limy	500 million+ AI bot visits over 90 days	408
OtterlyAI	62,100+ AI bot visits to one site, 90 days	84 (about 0.1%)
Ahrefs	137,210 domains in May 2026	Zero for 97% of files

None of these studies is perfect. Together, they cover one carefully built site, a large sample of real business sites, and a huge pool of bot traffic. All three point the same way.

Why we don't ship it

When we decide what goes into a Tilazo site, the test is simple: does it help a customer find you? On today's evidence llms.txt doesn't, so listing it as a feature would be selling you something that sounds useful and does nothing. We'd rather spend that effort on the parts AI crawlers do read.

If that changes, and a major AI search tool starts using the file, we'll look at it again. Adding one later is easy.

What to do instead

These are the things that move the needle, roughly in order. Our guide to how local businesses get found on ChatGPT walks through each one in detail.

1. Let the right crawlers in. Check yoursite.com/robots.txt. OpenAI says sites that block OAI-SearchBot "will not be shown in ChatGPT search answers." That one line matters far more than any summary file.
2. Put your facts in plain text. Services, service area, hours and phone number should be in the page itself, not only in images or loaded by scripts. Right-click your page, choose "View page source," and search for your phone number. If it isn't there, a crawler may not see it.
3. Give each service its own page. "Drain cleaning in Tacoma" on its own page is easier for an AI to quote than a bullet on your homepage.
4. Add LocalBusiness schema. It's a short block of code that labels your business type, address, hours and phone for machines. It removes guesswork.
5. Build reviews and listings. AI tools read what other sites say about you. Steady, detailed reviews and correct listings on Google, Bing, Apple Maps and Yelp do more than anything you put in a file. Our guide to getting more Google reviews covers how to ask.

If you already have one

Leave it. An llms.txt file doesn't hurt anything, and none of these studies found a downside. Just don't pay extra for one, and don't count it as your AI search plan.

Where Tilazo fits

Published Tilazo sites are built on the server, so the full text arrives in the first response. They include LocalBusiness schema and ship a robots.txt that allows OAI-SearchBot, PerplexityBot and the other major AI crawlers. That covers the first four items on the list above. Reviews and listings are still yours to build, and no website can do that part for you.

Want to see how your current site looks to a crawler? The free graded report checks your robots.txt for common AI crawlers, your schema, headings and page content, and asks Perplexity and Gemini the kinds of questions your customers ask.