Study Finds No Clear LLMs.txt Effect

New Analysis Shows Little Evidence That llms.txt Influences AI Citations Across 300,000 Domains

A recent study by SE Ranking has taken a close look at how widely the llms.txt file is being used and whether it has any effect on how often websites are cited by major AI models. After analysing roughly 300,000 domains, the company found no meaningful link between adding the file and securing more citations in large language model responses.

The research suggests that while interest in llms.txt is growing, the file is not yet shaping AI visibility in any measurable way.

What the Data Reveals

Adoption Remains Low

One of the clearest findings from SE Ranking’s review is that llms.txt has not been widely implemented. Out of all domains studied, only around one in ten had added the file to their root directory. This means most sites — nearly 90% — have not adopted it at all.

This low uptake is important because the format is sometimes positioned as an emerging standard for signalling AI permissions and preferred usage. Instead, the data shows that usage is still very experimental, with no strong pattern in who is adopting it.

Interestingly, adoption was not concentrated among large, high-authority websites. In fact, high-traffic domains were slightly less likely to use the file than mid-tier sites, suggesting that its uptake is currently scattered and inconsistent across the internet.

No Evidence of a Direct Impact on LLM Citations

Correlation Tests and Modelling

To determine whether llms.txt affects how often a domain appears within AI model responses, SE Ranking compared citation frequency across sites with and without the file.

The team used both statistical correlation tests and a more sophisticated XGBoost machine learning model. The intention was to see whether llms.txt played any measurable role in citation outcomes.

Surprisingly, the model performed better once the llms.txt input was removed. This means the file did not contribute meaningfully to predicting citation behaviour. SE Ranking concluded that there is currently no observable influence from the file on how often domains are referenced by LLMs.

Simpler statistical tools produced the same result: no significant correlation between citation frequency and the presence of llms.txt.

How This Aligns with Platform Guidance

Google’s Position

SE Ranking notes that these findings fit with what platforms have publicly stated so far. Google has never suggested that llms.txt is used as a signal for AI Overviews or Search’s AI Mode. In its documentation, Google explains that AI-generated answers continue to rely on existing Search systems, rather than introducing new ranking signals linked to llms.txt.

OpenAI’s Stance

Similarly, OpenAI’s published guidance focuses on robots.txt settings for managing crawler access. While GPTBot has occasionally been seen retrieving llms.txt files, OpenAI does not claim the file affects rankings, visibility, or citation patterns.

SE Ranking emphasises that although some logs show occasional llms.txt fetches, this behaviour appears inconsistent and does not correlate with increased AI citations.

What This Means for Website Owners

The main takeaway is that llms.txt is best viewed as an early-stage, optional experiment rather than a proven SEO or AI-visibility strategy. Adding the file is simple and generally harmless, so there is no major downside to implementation.

However, if your immediate goal is to improve visibility in AI-generated answers, the evidence suggests you should not expect a lift at this stage. The file is not currently used as a ranking or citation signal by major platforms, and adoption across the web is still limited.

Instead, llms.txt sits among a growing list of early tools designed for potential future interaction with AI systems. It may become more meaningful later if platforms begin adopting it formally, but for now it should be treated as a low-stakes test rather than a guaranteed advantage.

Conclusion

SE Ranking’s large-scale analysis makes it clear that, despite growing interest, llms.txt has no measurable impact on LLM citation behaviour today. Adoption remains low, platforms do not treat it as an input, and models do not appear to reward sites that use it. While adding the file may be useful preparation for future standards, it currently provides no proven visibility benefit in AI search or generative responses.