The llms.txt Truth: Cheap Insurance, Not a Strategy
Everyone is adding llms.txt. The honest read of the evidence: ship it, but do not expect it to get your reviews cited. Here is what actually does.
What is llms.txt, and what is it supposed to do?
An llms.txt file is a Markdown file you place at the root of your domain, at /llms.txt, listing your key pages with short descriptions. The idea, proposed in 2024, is to give large language models a clean, curated map of your site so they do not have to wade through navigation, scripts, and clutter to find what matters.
It is a sensible idea on paper. A buyer-facing site is full of noise a model has to discard. A short, honest index of your real pages is, in principle, a courtesy to any system trying to read you.
Does llms.txt actually get my reviews cited?
Not on the current evidence. The major answer engines have not confirmed that they read llms.txt, and the measurable effect on citation today is negligible. ChatGPT, Perplexity, and Google AI Overviews assemble answers from text they can extract and from sources they already trust, not from a file most of them have not committed to reading.
This is the part most write-ups skip. A file you can publish in five minutes is being sold as an AI visibility strategy. It is not one. Treating it as the work is how stores end up with a tidy /llms.txt and reviews that still never get quoted.
So should I ship llms.txt at all?
Yes, ship it. The case for llms.txt is not that it works today; it is that it is close to free and might matter later. Honest framing: it is insurance, not a strategy.
If adoption grows and engines start reading it, you are already covered. If they never do, you have lost a few minutes. Just do not let the easy task crowd out the work that moves citation now.
- It costs minutes to write and adds no risk to your site.
- It may pay off if engine support grows, and you lose little if it does not.
- It is a curated index, not a ranking lever, so judge it as housekeeping.
- It is no substitute for readable HTML, schema, and third-party profiles.
What actually gets my reviews cited instead?
Three things do the real work, and none of them is a single file. Answer engines quote text they can read, structured data they can parse, and sources they already trust. Your effort belongs there.
Most review apps were built for the on-page shopper and stop there, which is the exact gap BetterReviews is built to close: getting the reviews you already have readable, corroborated, and cited by the engines buyers now ask.
- Readable review HTML: reviews rendered into the server page, not locked inside a JavaScript widget the model sees as empty.
- Structured data: Product and Review schema that states ratings and review text in a form a machine can lift cleanly.
- Third-party profiles: populated listings on G2, Capterra, or Trustpilot, because an independent voice is what these systems prefer to cite.
Why does readable HTML beat a clever text file?
Because the model has to be able to read the review before any file can point it anywhere. Most review apps inject reviews through a JavaScript widget after the page loads. A shopper sees stars and quotes; a crawler often sees a placeholder where the social proof should be.
An llms.txt that links to a page whose reviews never render in the HTML points the model at an empty room. Fix the page first. The text has to exist in a form a machine can extract, or no map will help.
Is llms.txt the same as robots.txt or a sitemap?
No, and conflating them is a common mistake. A sitemap and robots.txt are widely supported standards that crawlers genuinely honour: one lists your URLs, the other states crawl rules. They have years of adoption behind them.
llms.txt is a proposal, not a ratified standard, and engine support is unconfirmed. Keep your sitemap and robots.txt correct first, because those are load-bearing today. Add llms.txt on top as a low-cost bet, not as a replacement for either.
What this adds up to
Ship llms.txt and forget about it. Spend the time it would have taken on a strategy elsewhere: get your review text into the server HTML, mark it up with clean schema, and build the third-party profiles a model can lean on. Those three are what decide whether an answer engine names you.
The honesty here is the point. Anyone selling llms.txt as the answer is selling you the easy half-hour and skipping the work. Do the cheap thing, then do the real thing.
- Does llms.txt help my reviews show up in ChatGPT?
- Not on the current evidence. Its measured effect on AI citation today is negligible, and major engines have not confirmed they read it. Ship it as cheap insurance, but get the real work done: readable review HTML, structured data, and third-party profiles do the lifting.
- If it does not work, why bother adding llms.txt?
- Because it is close to free and might matter later. It takes about five minutes, adds no risk, and covers you if engine support grows. The mistake is treating that five-minute task as a strategy rather than as insurance on top of the work that actually moves citation.
- Is llms.txt a real web standard like a sitemap?
- No. A sitemap and robots.txt are widely supported and genuinely honoured by crawlers. llms.txt is a 2024 proposal with unconfirmed engine support. Keep your sitemap and robots.txt correct first, and add llms.txt as a low-cost bet on top, not as a substitute.
- What should I do instead of relying on llms.txt?
- Make your reviews citable at the source. Render review text into the server HTML rather than a JavaScript widget, add Product and Review schema, and build third-party profiles on G2, Capterra, or Trustpilot. Closing that gap is what BetterReviews is built to do.