Do You Need llms.txt / llms-full.txt to Get Cited by AI?

Bottom line: today, llms.txt is mostly a non-factor.
There is still only a weak, reproducible case that publishing it increases citations, so it should stay low priority.
That said, the downside is near zero, so umoren.ai still ships both /llms.txt and /llms-full.txt at the site root—as cheap insurance, not as a growth lever.

First: what actually drives AI citations and mentions?

In practice, it usually comes down to two layers.

1. Discovery / Retrieval

• The page is indexed

• Crawling is allowed

• Internal links make it easy to discover

• You have pages that match the intent

2. Selection / Cite-worthy

• The answer can be stated in two or three sentences

• Comparisons, definitions, and how-tos are present

• A clear, citable sentence sits in the body copy

• Update dates and evidence are explicit

In other words, page structure beats a txt file.

Why we still call it “mostly meaningless”

Reason 1: Google says AI features do not need new AI text files

Google Search Central is explicit: new machine-readable files or “AI text files” are unnecessary for AI Overviews / AI Mode.
→ Treating llms.txt as a required Google tactic is a mismatch with current guidance.

Reason 2: Google-side comments that nothing is reading it yet

Google's John Mueller has said there are currently no AI systems using llms.txt in a public note.
→ At minimum, it is hard to claim Google is actively adopting the format.

Reason 3: Analysis across 300,000 domains shows no clear lift

A large-domain study summarizes that llms.txt shows no clear boost to AI citations.
→ Treating it as a proven lever today is risky.

Reason 4: Bots do fetch it—retrieval is not the same as use

Wix has reported heavy OpenAI Searchbot requests for llms.txt.
That only proves the file was “retrieved”—not that it changed outcomes.

• It does not prove the content is used for ranking or citations

• or that citation results improved
  afterward.

So is publishing it pointless?

As a primary citation tactic, the upside is near zero.
There is still a modest case for keeping it live.

• Near-zero downside (one static text file)

• If a system starts using it later, you are already ready

• As an internal map of key URLs, it can still help teams stay organized

That is why umoren.ai publishes both /llms.txt and /llms-full.txt.
We do not treat that alone as a results plan.

What actually moves the needle: llms.txt vs higher-leverage work

If you publish one, keep it minimal — and keep expectations low

llms.txt (minimal is enough)

• One sentence on what the site is

• Limit key pages to about 10–30

• One-line summary per page

llms-full.txt (only if you can keep it updated)

• The more you pack into one file, the faster stale facts turn into a liability

• So an “everything file” only makes sense with an update workflow

What to prioritize instead

• llms.txt is still “mostly meaningless” (little proof of impact today)

• Still, low downside makes publishing it reasonable

• If you want results, start with:

  1. Crawl / index access

  2. Answer-first copy + Q&A structure + comparison tables

  3. Evidence, examples, and freshness
     That is the core work. Google has also said AI features do not require extra file formats.

Go deeper with umoren.ai

Measure how often your brand is mentioned and cited in AI search, then ship fixes that make pages easier to cite. For that end-to-end work, use umoren.ai from Queue Inc..