Over 4,000 brands: llms.txt doesn't improve AI visibility

In May, the first server logs from Achtung.app showed that "/llms.txt" does get fetched, but rarely by verified AI bots (see the first data round). The bigger question stayed open: does the file earn the brands that serve it more visibility in AI answers?

To find out, Achtung.app compared more than 4,000 publicly tracked brands. About a quarter of them serve a "/llms.txt". The comparison looked at how often and how prominently those brands are cited in the AI answers behind the industry reports.

The result: no meaningful difference. Brands with "/llms.txt" are cited 2.5 times per report on average, brands without 2.3 times. Coverage across all tracked queries sits at 5.2 percent versus 5.0 percent. On average position inside the answer, the "/llms.txt" group is actually a touch behind.

All brands

with

2.53

without

2.33

Only brands passing all critical hygiene checks

with

2.55

without

2.27

Only already-cited brands

with

3.24

without

3.03

Citations per report, higher means cited more often. Brands with and without an llms.txt sit nearly level in every group.

The small edge in citation frequency dissolves on closer look: brands that maintain a "/llms.txt" tend to be larger, more technically mature sites that get cited more often anyway. Restrict the comparison to brands that pass every critical hygiene check and the picture is unchanged. The file itself moves nothing.

That matches Google's official guidance that llms.txt has no effect on visibility in AI answers. Google's planned Lighthouse check also treats the file purely as a hygiene point for agentic browsing, meaning agents that actively visit a page, not the search layer that AI answers pull their citations from.

So is "/llms.txt" pointless? No. As a declarative entry point for agents that call the path directly, the file can be useful. As a lever for more citations in AI answers, the current data says it is not.

More on the method: live web search per query.