llms.txt Won't Get You Cited by ChatGPT — Here's What Actually Will

If a web developer, SEO vendor, or AI consultant has pitched you this year on adding an "llms.txt" file to your website — the pitch usually goes something like: "It's the new robots.txt, except for AI. Add it, and ChatGPT, Gemini, and Perplexity will finally find and cite your business."

New server-log data says otherwise. Here's what's actually happening with llms.txt, and where your AEO budget should go instead.

The llms.txt Hype, By the Numbers

Content-verification company Originality.ai tracked more than 3 million websites over the past year, and the growth curve is real:

• llms.txt adoption is up 8.8x — from 4,088 sites in June 2025 to 36,120 by May 2026.
• llms-full.txt grew 107.1x (23 sites to 2,463).
• ai.txt grew 99.3x (4 sites to 397).
• Combined, roughly 39,000 sites now run one of these three formats, with plain llms.txt making up 92.7% of that total.

On paper, that's a genuine wave of adoption — the kind of growth curve that usually means something real is happening. It's exactly why vendors have been pitching it so hard.

The Catch: 97% of These Files Get Zero AI Requests

Here's the part the pitch usually leaves out. Ahrefs pulled actual server logs from 137,000 domains in May 2026, and the usage data tells a completely different story than the adoption data:

• 97% of llms.txt files received zero requests that month. Not low traffic — zero.
• Of the small sliver of traffic that did hit these files, AI retrieval bots made up just 1.1% of total requests.
• Training crawlers barely registered: GPTBot at 4.51%, ClaudeBot at 0.80%, DeepSeekBot at 0.02%.
• The biggest group actually requesting the file? SEO audit tools (21.7%), unidentified bots (14.9%), and general web crawlers (13.1%) — in plain English, other SEO software checking whether the file exists, not AI systems reading it for answers.
• Even Fortune 500 companies aren't bothering: only 7.4% have added llms.txt, compared with 92.8% that have standard robots.txt (ProGEO.ai, March 2026).

If you paid someone to add this file expecting it to move the needle on AI visibility, the honest answer is: almost certainly, it hasn't.

Why Isn't Anyone's AI Actually Reading It?

This isn't a mystery — the platforms themselves have said so directly:

• Platform guidance conflicts. OpenAI and Anthropic both point site owners back to standard robots.txt for managing crawler access. llms.txt shows up in their own developer documentation, not as an actual retrieval or ranking signal.
• Google has said it plainly. Google states llms.txt has no effect on Search rankings or on what shows up in AI Overviews.
• John Mueller put it bluntly in a widely-cited statement: "no AI system currently uses llms.txt."
• robots.txt can't even tell the crawlers apart. It can't distinguish a search crawler from a training crawler, and compliance with either file is entirely voluntary — there's no enforcement behind any of this.
• With no measurable traffic or citation benefit to point to, most site owners have no real reason to maintain the file long-term — which is a big part of why usage is flatlining even as sign-ups keep climbing.

So Should You Delete It? Not Necessarily

This doesn't mean llms.txt is a scam or that it's actively harmful to have. A few honest caveats:

• Researcher Mike King (iPullRank) argues that retrieval-first systems like Claude and Perplexity may use different architectures than a standard search crawler, and could plausibly read structured files like this differently in the future. Nobody's claiming this is permanently dead.
• The genuine use case today is narrow: sites with heavy documentation, public APIs, or large structured product catalogs — the kind of site where an AI agent (not a search engine) might need to navigate a lot of machine-readable content. Think a SaaS company's developer docs, not a local HVAC company's homepage.
• If you're a typical small business — a service company, a local retailer, a consultancy — having the file costs you almost nothing. But it is not where your next AEO dollar or your developer's next hour should go. Treat it as a "maybe eventually," never as a strategy.

What Actually Gets Small Businesses Cited by AI (Do This Instead)

This is the part that actually moves the needle, based on what's consistently confirmed to work across AEO and GEO studies this year:

• Structured data / schema markup on your real pages. FAQ schema, Organization schema, Service schema — this is the signal AI systems have repeatedly confirmed they do parse, because it's the same structured data search engines have relied on for over a decade. We covered the citation-tracking side of this in more depth here.
• Write in direct-answer format. Short, clearly-labeled answers to the exact questions your customers ask — not three paragraphs of throat-clearing before the actual answer shows up. Answer engines favor content shaped like the question itself.
• Keep your business details identical everywhere. Name, services, location, pricing ranges — your website, Google Business Profile, and every directory listing need to match. AI answer engines cross-reference these for confidence before they'll cite you, the same way they'd flag a ghost citation when the details don't line up.
• Publish something that isn't already a commodity. Original data, a real case study, your own numbers from your own business. AI engines are far more likely to cite a source that adds new information than one that restates what every competitor's site already says.
• Check what's actually measurable. Google Search Console's newer AI performance reports show which of your pages are already showing up in AI Overviews — that's real, trackable data, unlike a file almost nobody's AI is requesting. We walked through reading that report here.

The Bigger Pattern: Don't Chase the File, Chase the Fundamentals

SEO trackers are flagging renewed Google ranking volatility again this week, part of a pattern of spikes that's continued on and off since September's spam update — which we also broke down in terms of what actually changed for AI-assisted content. Different story, same lesson: chasing a trendy, one-time technical add-on is a worse long-term bet than fixing the fundamentals that hold up through every single update — accurate structured data, genuinely useful content, and consistent business information across every platform where a customer or an AI system might encounter you.

FAQ

What is llms.txt?
A plain-text file, modeled loosely on robots.txt, that site owners can add to tell AI systems what content on their site is okay to use for training or retrieval. It has never been adopted as a formal standard by any major AI company.

Does llms.txt actually help with ChatGPT or Google AI Overviews?
Not based on current data. Google has said directly it has no ranking effect, and server-log data shows under 1.1% of a typical site's traffic to that file comes from recognized AI bots — most hits are other SEO tools simply checking whether the file exists.

Should my small business add llms.txt anyway?
It won't hurt to have one, but it's a low-priority task for most small businesses. Sites with heavy documentation, public APIs, or large structured catalogs might see value eventually; a typical local service business gets far more AI-citation value from schema markup and direct-answer content today.

What should I actually do to get cited by AI search tools?
Add proper schema markup to your key pages, write FAQ-style direct answers to real customer questions, keep your business details consistent across every platform you're listed on, and check Search Console's AI performance reports to see what's already working before you add more.

What To Actually Do This Week

• Stop paying for or prioritizing an llms.txt file as an AI-visibility fix — it's a near-zero-ROI line item for most small businesses right now.
• Add or audit FAQ and Organization schema markup on your top five pages instead.
• Pull up your Google Business Profile, website, and top three directory listings side by side and fix any mismatched name, service, or location details.
• Check Search Console's AI/Generative performance report this week to see what's already showing up in AI Overviews.
• Write one piece of genuinely original content — a real number, a real case study — instead of another generic explainer.

The Bottom Line

llms.txt isn't a scam, and it isn't completely worthless — it's just not where the ROI is for a small business today, no matter how fast the adoption curve looks. The businesses actually getting cited by ChatGPT, Gemini, and Perplexity right now are the ones doing the less exciting, more durable work: clean structured data, direct answers, and consistent information everywhere an AI system might look.

This is exactly the kind of AEO/GEO audit cCoreVentures' AI Strategy & Implementation practice runs for small business clients — separating the trendy file from the fundamentals that actually get you cited, and building a plan around what the data shows instead of what a vendor is selling. If you've already paid for an llms.txt file and want to know what it's actually doing for you, let's find out together.