llms.txt Files: Do They Actually Help AI Visibility?
The llms.txt standard promises to help AI models understand your site. We tested it across ChatGPT, Perplexity, and Gemini. Here's what actually happened.
Marcus Torres
March 12, 2026
What is llms.txt?
llms.txt is a proposed standard (similar to robots.txt) that lets website owners provide a structured, markdown-formatted summary of their site's content specifically for AI models. The idea is that instead of AI systems crawling and parsing your entire site, they can read a concise, curated summary of who you are and what you offer.
Our Test Methodology
We implemented llms.txt on 50 websites across different industries and monitored ChatGPT, Perplexity, and Gemini citation rates over 12 weeks, compared against a matched control group without llms.txt. Prompts were standardized across both groups.
Results: What We Found
Perplexity: Statistically significant improvement (+8.3% citation rate) for sites with well-structured llms.txt files. Perplexity explicitly reads llms.txt during its crawling process.
ChatGPT browsing mode: Modest improvement (+3.1%) — ChatGPT appears to occasionally read llms.txt during web browsing sessions.
ChatGPT base model: No measurable impact — base model responses don't use live web data.
Gemini: No measurable impact detected in our test window. Google hasn't confirmed llms.txt support.
Bottom line: llms.txt is worth implementing (it takes ~30 minutes), primarily for its Perplexity benefit. Don't expect it to be transformative — it's a marginal improvement on top of a strong content and schema foundation.
How to Implement llms.txt
1. Create a file at `yourdomain.com/llms.txt` 2. Include: your organization name and one-line description, key product/service descriptions, your target audience, links to your most important pages, and a brief FAQ in markdown format 3. Keep it under 10,000 tokens (roughly 7,500 words) 4. Submit the URL to Perplexity's crawler via their webmaster tools 5. Update it quarterly when your product or content changes significantly
Frequently Asked Questions
Is llms.txt an official standard?
No — llms.txt was proposed by Answer.AI (Jeremy Howard) but is not an official W3C or IETF standard. However, Perplexity has stated they support reading it, and several major AI platforms are watching its adoption. It's low-risk to implement now.
Should I create both llms.txt and llms-full.txt?
The llms.txt standard proposes a short summary file and an optional llms-full.txt with complete content. Implementing the main llms.txt is sufficient for most sites. Only add llms-full.txt if your product is highly technical and would benefit from AI systems having detailed specifications.
Track your AI search visibility with amplerank
Monitor your citation rate across ChatGPT, Perplexity, and Gemini — and see how you compare to competitors.
Get a free demo