llms.txt for a local business is a short markdown file at your site's root that lists your services, service area, contact details, and key pages in a clean format that AI tools can read easily. It takes about an hour to write, costs nothing to host, and is a sensible extra for a local business site. It is not a replacement for your Google Business Profile, your reviews, or your local citations, and any guide that promises otherwise is overselling it.
The honest starting point is the evidence. Several independent log studies in 2026 found that AI crawlers almost never request llms.txt. In one study covering more than a thousand AI crawler visits, robots.txt was requested hundreds of times, sitemaps dozens of times, and llms.txt not once. Other studies found the same pattern: most llms.txt files on the web receive zero requests. Google's John Mueller has also said that no AI system currently uses llms.txt. That does not make the file useless, but it does set the priority. Build it after your core local SEO is solid, not before.
What llms.txt actually is
llms.txt is a proposed convention, not a formal standard. A site places a markdown file at /llms.txt, with a title, a short description, and sections listing its main pages with brief summaries. An optional larger version, llms-full.txt, can hold the full text of key pages. The idea is that an AI tool can read one file instead of crawling dozens of pages to understand what a site offers.
For a local business, that idea translates into something simple: one page that states, in plain language, what you do, where you do it, who you serve, and where to find details. A plumber's file might list emergency repairs, boiler service, and bathroom fitting, with the towns covered and a link to the booking page. A dentist's file might list check-ups, whitening, and emergency appointments, with opening hours and the new-patient page. Nothing in that list is new information. It is your existing site, condensed.
What to put in a local business llms.txt file
Start with identity and service area
Open with your business name, what you do in one sentence, and the towns or neighbourhoods you serve. This is the part an AI answer needs most, because local questions are usually phrased as a service plus a place: a plumber near a town, a dentist open on Saturday, a restaurant that takes bookings. If your service area is vague, the file cannot fix that. Name the places you actually serve.
List services with one line each
List each service as a link to its page, followed by a single plain sentence about what it includes. Keep it to your real services. A file that lists twenty services you barely offer will confuse readers and AI tools alike, and it will not match the pages the links point to. Five to ten genuine services, well described, beat a long invented list.
Add the facts customers ask about
Include opening hours, phone number, address, booking link, and any facts that decide a booking: emergency call-out availability, parking, wheelchair access, languages spoken, or whether you take new patients. These are the details that turn a search into a visit, and they are often buried three clicks deep on a business site. Put them in the file where they can be found in one read.
Link to proof, not just claims
Link to your reviews page, your price or FAQ page, and two or three of your best service pages. If you have before-and-after photos, case studies, or a clear pricing page, link those too. AI answers tend to favour pages that show evidence, and a file that only claims quality without pointing to proof is weak. Do not invent reviews, awards, or statistics to fill the file. If a claim cannot be verified on your site, leave it out.
Keep it short and current
A local business llms.txt file should be short enough to read in two minutes. Aim for a few hundred words, not a few thousand. And update it when the business changes: new services, new hours, a moved premises, a closed branch. An out-of-date file is worse than no file, because it gives a confident wrong answer to anyone who reads it.
An example structure
A practical structure for a local business looks like this, in plain markdown:
# Business Name — Service in Town > A one-sentence description: what you do and where. ## Services - [Service one](https://www.example.com/service-one): one plain sentence about it. - [Service two](https://www.example.com/service-two): one plain sentence about it. ## Service area - Town one, town two, and nearby villages, with any limits stated plainly. ## Visit and contact - Address, phone, opening hours, and booking link. ## Useful pages - [Prices](https://www.example.com/prices) - [Reviews](https://www.example.com/reviews) - [Frequently asked questions](https://www.example.com/faq)
Generate a draft with a free llms.txt generator, then edit the draft by hand. Generators are good for structure and formatting, but they cannot know your real service area, your real hours, or which services you actually want more work from. The hand edit is where the value is.
What llms.txt cannot do for a local business
First, it cannot create local visibility by itself. A business with no Google Business Profile, no reviews, and no local citations will not start appearing in AI answers because it added a text file. AI answers draw on the same local signals search engines have always used: a complete and accurate business profile, consistent name and address details across the web, genuine reviews, and pages that match local queries.
Second, it cannot control AI training. That job belongs to robots.txt, which names crawlers such as GPTBot and ClaudeBot and tells them not to use content for training. llms.txt is a discovery file, not a permission file. Do not confuse the two, and do not rely on llms.txt to opt out of anything.
Third, it cannot replace your website. If your service pages are thin, your prices are missing, and your contact page is out of date, a summary file that points to those pages will not help. Fix the pages first, then summarise them.
The right priority order for local AI visibility
If you have one afternoon to improve how AI tools see your local business, spend it in this order:
- Google Business Profile. Complete every field, add real photos, keep hours accurate, and answer reviews. This is still the strongest local signal.
- Your own service pages. One clear page per service, with the town names customers actually search for, prices or price ranges where possible, and a booking step that works.
- Consistent business details. The same name, address, and phone number on your site, your profile, and your main directories. Inconsistency here creates doubt in every system that reads it.
- Robots.txt decisions. Decide which AI crawlers may train on your content and which search bots you want to allow, and write those rules in robots.txt. A robots.txt generator keeps the syntax clean.
- llms.txt as the finishing step. Once the above are solid, add the file as a tidy summary. It costs little, it may help future tools, and it forces you to write down your offer clearly, which is useful in itself.
That order is the honest one. A guide that puts llms.txt first is selling a file, not a local visibility strategy.
Should every local business add llms.txt?
Most should, eventually, because the cost is low and the downside is near zero. The exceptions are businesses with no website maintenance at all: if nobody will update the file when hours or services change, skip it until someone owns it. An accurate file that is a year out of date is a liability, and a small business with a single-page site may get more value from improving that page than from adding a second file nobody maintains.
Also skip the full version, llms-full.txt, unless you have a clear reason for it. For most local businesses, the short file is enough. The full version duplicates your whole site into one file, which is harder to keep current and adds little for a business with ten pages.
Measuring whether it helps
Do not expect a traffic jump from llms.txt, and do not judge it by one. Check your server logs occasionally for requests to the file. If you see none over a few months, that matches the published studies, and it tells you the file is a future bet, not a current channel. Keep it because it is cheap and accurate, not because it is driving bookings today.
Judge your local AI visibility instead by the things that do move: whether your business appears in AI answers for service-plus-place queries, whether your profile and reviews are complete, and whether enquiries mention AI tools as the source. If those improve, your core local work is paying off. The llms.txt file is a small part of that story, and it is fine for it to stay small.
FAQ
Does llms.txt help a local business rank in AI answers?
There is no solid evidence that it does today. Log studies show AI crawlers rarely request the file, and Google has said no AI system currently uses it. Treat it as a low-cost extra, not as a ranking tactic.
Is llms.txt a replacement for robots.txt?
No. Robots.txt controls crawling and training opt-outs. llms.txt summarises your site for AI tools. They do different jobs, and a local business that wants both needs both files.
Where does llms.txt go on a website?
At the root of your domain, at yoursite.com/llms.txt, in markdown format. Most hosts and builders let you upload a file to the root, or you can place it in the static files folder your platform provides.
How long should a local business llms.txt file be?
A few hundred words is enough: identity, service area, services with one line each, contact details, hours, and links to key pages. Longer files are harder to keep accurate and add little for a small business.
Do I need llms-full.txt too?
Usually not. The short file covers a local business well. Add the full version only if you have long documentation or many pages that a summary cannot fairly represent.
What matters more than llms.txt for local AI visibility?
A complete Google Business Profile, genuine reviews, consistent business details across the web, and clear service pages with local place names. Those signals drive local answers today. llms.txt sits behind all of them in priority.
