If you have talked to anyone about AI search in the last year, someone has probably told you to add an llms.txt file to your website. The pitch is appealing: one small text file, and suddenly ChatGPT, Claude, and Perplexity understand your business better and start recommending you.
At Mile High Content, a boutique digital marketing agency in Denver, Colorado, we looked at the research. Do local businesses actually need an LLMS.TXT file? Here is the honest version.
WHAT LLMS.TXT ACTUALLY IS
llms.txt is a plain text file that lives at the root of your domain, at yoursite.com/llms.txt. Developer Jeremy Howard proposed it in September 2024.
The file is written in Markdown, a simple formatting style that machines read easily. Inside, you write a short description of what your business does, followed by a curated list of your most important pages, each with a one-line explanation. A version of your site, stripped of navigation menus, popups, and ads, handed directly to an AI system.
The idea makes sense. AI models operate within a limited reading space. If a model has to wade through your entire website to figure out what you do, it may get it wrong. A curated file tells it exactly where to look.
LLMS.TXT IS NOT WHAT YOU THINK IT IS
Two corrections are worth making, because both circulate widely.
It is not “robots.txt for AI.” Robots.txt controls access. It tells crawlers what they may and may not visit. llms.txt does not block or grant anything. It is a recommendation, not a rule. If you want to control which AI bots crawl your site, robots.txt is still the file to use.
It is not a ranking factor. No search engine or AI platform uses it as a scoring signal. Anyone selling it as a shortcut to AI visibility is ahead of the evidence.
WHAT THE DATA SAYS ABOUT LLMS.TXT
This is where the story turns.
SE Ranking analyzed roughly 300,000 domains to test whether the presence of an llms.txt file correlates with how often a site gets cited in AI answers. They found no statistically significant relationship. When they removed llms.txt from their prediction model entirely, the model got more accurate. The file was adding noise, not signal.
The same study found adoption sitting at about 10 percent of domains, spread fairly evenly across traffic tiers. Among the 50 most-cited domains in AI answers, exactly one had an llms.txt file.
Crawler logs tell a similar story. In one published 90-day study of AI bot traffic, requests to llms.txt accounted for roughly 0.01% of all AI crawler activity on the monitored site.
Google has addressed it directly. John Mueller compared llms.txt to the old keywords meta tag, noting that no AI service has said it uses the file and that server logs show the bots do not even request it. His concern is the same one that killed the keywords tag: a site owner claiming what a site is about without verification. If a machine can simply read the site, why trust a summary the site wrote about itself?
SO WHY DO SMART COMPANIES STILL PUBLISH ONE?
Look at who has adopted it: Anthropic, Stripe, Cloudflare, Vercel, Supabase, and LangChain. Almost without exception, developer infrastructure companies with large technical documentation libraries.
That pattern is the answer. For those companies, the file has a real, present-day use. Coding assistants and developer tools genuinely fetch llms.txt to pull documentation into a workspace. The audience is engineers using AI tools to read technical docs, not customers asking an AI for a recommendation.
There is also a reasonable bet being placed. The file takes a few hours to build. If a major AI platform decides tomorrow to start honoring it, the sites that already have one are simply ready. Low cost, unknown payoff, no downside. For a company with a dedicated web team, that math works fine.
OUR RECOMMENDATION FOR LOCAL BUSINESSES
If you run a restaurant, a dental practice, a law firm, or another local business in the United States, here is our position:
Publish an llms.txt file if it is free. Skip it if it is not.
If your CMS or WordPress plugin generates the file automatically, turn it on, check that the output looks clean, and move on with your day. Zero cost, zero risk, and you are positioned if the standard gains traction.
If building one means paying a developer or pulling hours away from other marketing work, that is the wrong trade right now. The evidence does not support it, and your budget has better places to go.
We would rather tell you that than sell you a file.
WHERE THE AI VISIBILITY WORK ACTUALLY LIVES
AI systems are not ignoring your business. They are reading your website, your Google Business Profile, your reviews, and what other sites say about you. That is where the work belongs.
Your Google Business Profile. For local searches, this remains the single most consulted source of truth about your business. Categories, hours, service areas, photos, and the Q&A section all feed how AI systems describe you.
Structured data. Schema markup is the version of “explaining your site to machines” that platforms actually read and act on. LocalBusiness, FAQPage, Product, and Review schemas convey the same information that llms.txt aims to, via a channel with confirmed support.
Direct answers on the page. AI systems quote sentences, not pages. Content that answers a question clearly, in a complete sentence, near the top of the page, gets pulled into answers. Content that buries the answer in paragraph nine does not. YouTube videos help support this evidence, especially when embedded on relevant website pages.
Consistency everywhere. Your name, address, and phone number should match across every directory, profile, and citation. Conflicting information makes an AI system uncertain about you, and that uncertainty leads it to name a competitor instead.
Reviews and mentions. What third parties say about you carries weight that self-description never will. This is precisely why Mueller’s keywords-tag comparison stings: outside verification beats self-report every time.
None of that is new. It is the same local search foundation that has driven local SEO results for years, now serving a second audience through AI.
THE TAKEAWAY
llms.txt is a well-intentioned proposal that has not yet earned its place in a small business marketing plan. It might. The people behind it are serious, and the problem it addresses is real. But today it is a bet, not a tactic, and we do not spend client budget on bets when the fundamentals are still available.
If you want your business to show up when a customer asks an AI where to eat, who to hire, or which local company to trust, the path runs through your Google Business Profile, your structured data, your reviews, and content that answers real questions clearly.
That work is what we do. Mile High Content builds local search and SEO strategies for small- to medium-sized businesses, and we work with a limited client roster, so you deal with senior specialists rather than an account coordinator. If you want to know where your business currently stands with both search engines and AI systems, let us take a look.
