Is llms.txt / llms-full.txt necessary for citation by AI?

Conclusion: At this point, it is almost meaningless.
At least, there is a weak reproducible basis for "the increase in citations due to this." Therefore, its priority is low.
However, since there is almost no harm, Umoren.ai places both /llms.txt and /llms-full.txt at the root directory. If you're going to do it, consider it as "insurance."

First, let's clarify: What determines whether AI cites or mentions something?

Whether AI picks it up is roughly determined by these two factors.

1. Discovery / Retrieval

• Indexed

• Crawling is allowed

• Easy to follow through internal links

• Pages that match search intent exist

2. Selection / Cite-worthy

• Conclusions can be stated in 2-3 sentences

• Comparisons, definitions, and procedures are provided

• A "citable sentence" is embedded in the text

• Update dates and basis are clear

In other words, the structure of the text takes precedence over the txt file.

The basis for saying "almost meaningless"

Basis 1: Google explicitly states "No new AI text is needed for AI features"

Google Search Central clearly states that new machine-readable files or “AI text files” are unnecessary to appear in AI Overviews / AI Mode.
→ Treating llms.txt as a "mandatory measure for Google" is currently misaligned.

Basis 2: Reports of "no one is reading it" (as Google's perspective)

John Mueller from Google has stated that "currently, there are no AI systems using llms.txt" in a statement.
→ At least, it is hard to say that "Google is actively adopting this specification."

Basis 3: Data showing "no clear effect from analyzing 300,000 sites"

As a large-scale domain analysis, there is a summary stating that there is no evidence that llms.txt clearly boosts AI citations available.
→ At this point, treating it as a "certainly effective measure" is risky.

Basis 4: On the other hand, there are "observations" that bots are coming to retrieve it (which does not guarantee they are reading it)

According to Wix's observations, there have been reports of the OpenAI Searchbot coming to retrieve llms.txt in large quantities as a case.
However, this only means they “retrieved it”,

• but it does not prove that they are using its content for ranking/citations

• or that the results have improved
  as a result.

So, does it have zero meaning?

As a "main measure to increase citations," it is close to zero.
However, in this sense, there is "value in keeping it."

• Almost no harm (just adding one static text)

• If somewhere starts referencing it in the future, having it in place means immediate response

• As a "list of important pages" for site management, it can be used for internal organization

Therefore, Umoren.ai places both /llms.txt and /llms-full.txt.
However, we do not assume that results will come just because we did this.

Which is effective? Comparison table of "llms.txt vs truly effective measures"

If you're going to place it, this is all you need (minimum configuration) — but don't expect much

llms.txt (minimum is fine)

• A one-sentence description of the site (what kind of site it is)

• Narrow down important pages to 10-30

• One-line descriptions for each page (what is written there)

llms-full.txt (only if you have a "system that can be updated")

• If you cram too much into one file, the moment old information mixes in, it can easily become counterproductive

• Therefore, if you are going to create an "all-in-one," it should include an update flow

Finally: The real priority of what should be done

• llms.txt is "almost meaningless" (there is little certainty of its effectiveness at this point)

• However, since there is little harm, it is acceptable to keep it

• To achieve results, first

  1. Crawling / Indexing

  2. Presenting conclusions first + Q&A structure + comparison tables

  3. Evidence, examples, and updatability
     This is the main point. Google has also stated that no additional requirements are needed for AI features.

You can do more specific things with Umoren.ai

You can quantify (visualize) how much your company is mentioned and cited in AI searches, and implement improvements to shape it into a citable form. For more specific tasks, you can use Umoren.ai from Queue Corporation.