Most HubSpot customers do not need an llms.txt file today. For most marketing websites, improving content quality, structured data, CRM accuracy, and crawlability will have a much greater impact on AI visibility than implementing llms.txt.

If you’ve read anything about AI search in the past year, you’ve probably seen llms.txt mentioned as a must-have, a new file, sitting at the root of your domain, that supposedly tells AI models how to understand and cite your site. It gets pitched with the same urgency robots.txt had in the early 2000s.

The honest answer is more nuanced than that pitch suggests, and it’s worth understanding before you spend a sprint on it. Here’s what the current data actually shows, and where llms.txt does and doesn’t fit into a HubSpot marketing team’s priorities.

What llms.txt Is

llms.txt is a plain-text markdown file, proposed by data scientist Jeremy Howard in September 2024, that sits at a website’s root (yourdomain.com/llms.txt) and gives AI systems a curated summary of a site’s purpose and most important pages. The idea behind it is straightforward: large language models have limited context windows and struggle to process an entire website at once, so a short, curated index is meant to make a site easier to navigate. For most HubSpot-managed marketing websites, llms.txt should be viewed as an optional enhancement rather than a foundational SEO or AEO requirement..

Two related formats have emerged alongside it. llms-full.txt takes the same concept further, flattening an entire site’s content into one large markdown document rather than just linking outward. ai.txt, promoted by a company called Spawning, works differently, instead of helping AI models navigate a site, it issues directives about what AI crawlers are and aren’t allowed to do with the content, including training-use restrictions and licensing terms.

The Adoption Numbers

Adoption has grown quickly in relative terms and remains small in absolute terms. Tracking research from Originality.ai, which monitors more than 3 million websites, found llms.txt implementations climbing from roughly 4,000 sites in June 2025 to over 36,000 by May 2026, nearly a ninefold increase in a year. Separate research from a top-1,000-domain tracker puts current adoption among the world’s largest sites at under 9%, and a 300,000-domain study by SE Ranking found overall adoption sitting just above 10%.

Adoption skews toward specific sectors, too. A study of Fortune 500 companies found only about 7% had implemented an llms.txt file as of early 2026, a stark contrast to robots.txt, which nearly 93% of the same companies had in place. Adoption has been strongest among developer-facing SaaS companies, documentation-heavy platforms, and technically sophisticated brands; it remains rare in regulated industries like financial services, healthcare, and legal.

The Part Most Coverage Leaves Out: Almost No One Is Reading These Files

Here’s the finding that actually matters for a HubSpot marketing team deciding where to spend time: adoption and usage are two very different numbers, and the gap between them is enormous.

Server-log research from Ahrefs, covering 137,000 domains, found that 97% of llms.txt files received zero requests in a recent monthly period. Where AI systems did request the file, the traffic wasn’t coming from the tools most marketers care about. AI assistants that answer user queries, the ChatGPTs and Perplexitys of the world, accounted for just over 1% of requests. Training crawlers fared a little better individually: OpenAI’s GPTBot represented about 4.5% of requests, and Anthropic’s ClaudeBot less than 1%. The largest single category of requester wasn’t an AI system at all, it was SEO audit tools, checking whether the file existed, not reading it for content.

Google has been unusually direct about where llms.txt stands for search visibility. In guidance published in mid-2026, Google stated plainly that the file has no positive or negative effect on search rankings and isn’t required. OpenAI and Anthropic‘s own crawler documentation points site owners toward robots.txt for managing AI crawler access, mentioning llms.txt only in developer-facing docs, even though both companies publish an llms.txt file for their own documentation sites.

An analysis run by SE Ranking using a machine learning model tested whether having an llms.txt file predicted AI citation frequency. The result: removing the llms.txt variable from the model actually improved its accuracy, meaning the file’s presence added noise rather than a real signal.

So Is It Worth Doing?

Based on where the data actually points, here’s the honest, no-hype recommendation:

Skip it if you’re a typical B2B or B2C marketing site, service pages, a blog, case studies, a pricing page. For example, organizations using HubSpot CRM, Marketing Hub, and Content Hub should prioritize clean CRM properties, refreshed content, and structured pages before considering llms.txt. There’s no evidence it improves whether AI Overviews or chatbots cite you, and your time is better spent on the fundamentals that do have evidence behind them: clean, well-structured content; accurate schema markup; strong topical authority; a healthy robots.txt that isn’t accidentally blocking AI crawlers you want to allow. 

Consider it if your site is documentation-heavy, API-driven, or built around structured technical content that AI coding assistants or agentic tools might need to navigate, think of a software product with public developer docs, a knowledge base, or a platform other tools integrate with. In that narrower case, some retrieval architectures (particularly coding assistants and documentation-aware agents) may use it, and the cost of implementing it is genuinely low.

Don’t treat it as a substitute for the content and structural work that reliably drives AI citations. If your CRM data is inconsistent, your landing pages are stale, or your content isn’t written in clear, retrievable chunks, an llms.txt file sitting on top of that won’t fix it. It’s a low-cost, low-risk addition for the right kind of site, not a shortcut around the harder, higher-leverage work.

The Biggest Takeaway

llms.txt isn’t a scam and it isn’t nothing, it’s a young, unevenly supported standard with a narrow, legitimate use case that doesn’t match how most marketing teams are being told to use it. If someone’s pitching it to you as the key to AI search visibility, that’s a signal to ask more questions, not fewer.

If you’re trying to sort out which AEO tactics are actually worth your team’s time versus which ones are noise, that’s exactly the kind of prioritization conversation we help HubSpot marketing teams work through. Happy to talk it through with you.

If you want to see exactly where your brand stands in AI-generated answers right now, HubSpot’s AEO tool (currently in active beta, tied to Marketing Hub Pro/Enterprise) tracks your visibility across ChatGPT, Gemini, and Perplexity and turns the gaps into a prioritized action list.

This post contains an affiliate link. If you sign up through it, we may earn a commission at no extra cost to you.