What llms.txt can still do if Google ignores it
Useful for humans, weak for citations

What does llms.txt still do in practice?

The honest answer is boring, which is usually where the useful advice lives. llms.txt can still help you organise what you want machines and humans to find first. It can point to canonical product pages, policy pages, docs, glossaries, methodology pages, and evidence pages in one predictable place.

That matters inside a real company because AI visibility problems are rarely caused by one missing file. More often, the site says three different things in three different places, the best explanation sits in a PDF, the product page makes claims without support, and the changelog is impossible to parse. A file that forces you to choose the pages you actually stand behind can improve that mess.

• It gives your team a single shortlist of pages worth maintaining
• It exposes duplication and contradiction fast
• It can help external vendors understand your preferred source documents
• It creates pressure to publish missing pages, such as methodology or definitions
• It is easy to version, review, and update when positioning changes

What can llms.txt not do?

It cannot make weak content citable. It cannot override blocked crawling. It cannot rescue client side rendering when crawlers fetch JavaScript files and never run them. If the answer is not present in server delivered HTML, extractable headings, tables, lists, and plain claims with support, the file does not save you.

It also does not create trust by itself. A model or assistant looking for a source still has to decide whether your page is specific, stable, and worth quoting. If your claims are vague, self congratulatory, or unsupported, a neatly written llms.txt file just points systems toward weak material faster.

This is where a lot of teams go wrong. They buy the idea that one declaration file will influence discovery. The verified picture is narrower. Google states llms.txt is not used by Search. A study across about 300,000 domains found 10.13% adoption, none among the top 1,000 sites, and no citation lift after controls. That does not mean nobody reads the file. It means you should not expect search visibility gains from publishing it alone.

So why publish llms.txt anyway?

Because some work is still worth doing even when it is not a ranking factor. I would publish llms.txt if it helps your team operate better. That means clearer ownership, fewer stale pages being treated as canonical, and a better list of source documents for anyone evaluating your company, including AI systems, analysts, partners, and prospects.

For founder led companies and technical teams, the file can become a forcing function. You decide which pages express the business accurately. You remove old landing pages from that shortlist. You add definitions, methodology, limitations, and product boundaries. In other words, llms.txt can improve the quality of the pages around it, even if the file itself is not the reason you get cited.

There is also a simple communication benefit. When someone asks, what should an assistant read to describe us correctly, you have an answer. Not a pitch deck. Not a random Notion page. A maintained public list.

The operator use case I actually like

Use llms.txt as an editorial contract. Every URL in it should meet a higher bar than the rest of the site. It should load with meaningful server rendered HTML, answer a narrow question clearly, state definitions consistently, and show the source of any important claim. That turns the file into a shortlist of pages that deserve maintenance priority.

When does publishing llms.txt waste time?

It wastes time when the underlying site is the real problem. If your core pages are rendered in the browser, if facts are hidden in tabs that do not exist in HTML, if product names change page to page, or if the best explanation sits behind a form, your effort belongs elsewhere first.

It also wastes time when teams use it as symbolic compliance. I have seen this pattern with schema too. People publish a file or add markup because the market says AI optimisation is happening there, while the page itself remains hard to quote. Machines cite clean source material more readily than they reward decorative signalling.

• Do not start with llms.txt if your site relies on client side rendering for core copy
• Do not start with llms.txt if your claims have no evidence pages behind them
• Do not start with llms.txt if your brand, product, and category language are inconsistent
• Do not start with llms.txt if internal owners cannot keep the linked pages current

What should go inside llms.txt if you publish it?

Keep it strict. The temptation is to dump every commercial page into the file. That weakens the point. A short, opinionated list is more useful than a directory of everything you sell.

This is also where trade offs matter. If your business changes often, the maintenance burden is real. A stale llms.txt file can become a liability because it keeps pointing people and systems to pages you no longer endorse. In that case, skip it until someone owns it.

How does llms.txt fit with the work that actually moves AI visibility?

It sits downstream of the fundamentals. First make pages accessible to crawlers. Then make them extractable. Then make them worth citing. Only after that does llms.txt become a tidy wrapper around work you already did.

That means server delivered content, obvious page purpose, precise headings, direct answers near the top, stable definitions, and support for important claims. If you want the practical framework for that, start with extractability and citation mechanics, not declarations.

If your issue is demand generation or outbound execution rather than AI search visibility, that belongs with our parent company. We run managed outbound under Outbound Pros. Different problem, different operating model. Do not force llms.txt into a pipeline issue it cannot solve.

Who should publish llms.txt, and who should skip it?

Publish it if you have a real content estate, a clear point of view, and someone who can maintain a small set of authoritative URLs. It is especially sensible for documentation heavy companies, technical products, research led teams, and founder led businesses that care about exact wording.

Skip it, for now, if your site is small, your pages change weekly, or your main problem is basic discoverability and rendering. In those cases, llms.txt is housekeeping. Useful later, not urgent now.

That is the trade off in plain language. llms.txt can help you be easier to understand. It does not help you deserve attention. The second part still comes from better pages.