# Schema markup and llms.txt for AI search: what's actually proven

> In short: No. Google says structured data isn't required for its AI features and that Google Search doesn't use llms.txt or other AI text files. Keep the structured data you already use for rich results, make sure it matches what the page says, and add an llms.txt only if you want AI agents to find their way around your site or docs. Neither replaces a clear page.

Guide from Keats, published 2026-11-13, updated 2026-11-13. Canonical: https://getkeats.com/guides/schema-markup-ai-search

Two pieces of advice circulate in every AEO thread: "add schema so AI understands you" and "add an llms.txt file". Both sound plausible, both are cheap, and both are oversold. Here's what the evidence and the platforms actually say.

## What Google says about structured data

Google's guide to its generative AI features is direct: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." Its AI features page asks for one thing about the markup you do have: make sure your structured data matches the visible text on the page.

That's consistent with how these features work. AI Overviews and AI Mode are built on Google's Search ranking and quality systems and read pages much as Search does. Structured data still earns rich results in ordinary search (product prices and ratings, events, recipes), and those are worth having for their own sake.

## What about ChatGPT, Perplexity and the rest?

None of the major assistants documents structured data as a factor in which pages they cite. They read the page's text. Markup can't hurt if it's accurate, and it may help a crawler identify what a page is (a product, an article, an organization). But there's no published evidence that adding schema makes an assistant name you more often, and we don't claim any.

## The structured data worth having

If you're doing it anyway, these are the types with clear uses:

| Type | What it's for | Notes |
|---|---|---|
| Organization | Your name, logo, official site and profiles | Helps disambiguate you from similarly named companies |
| Product with Offer | Price, availability, reviews | Must match the price on the page |
| Article | Author, dates | Keep `dateModified` honest |
| FAQPage | Question-and-answer pages | Use only where the page really is an FAQ |
| LocalBusiness | Address, hours, area served | Pair with an accurate Business Profile |

The rule that matters more than the list: markup that disagrees with the page is worse than none. A Product block with last year's price is a wrong fact you've stated twice.

## What llms.txt is

An [llms.txt file](https://getkeats.com/glossary/llms-txt) is a markdown file at the root of a site that gives AI agents a short map: what the site is, and links to the pages or docs that matter, ideally in a clean text form. Jeremy Howard proposed it at llmstxt.org in September 2024, aimed at helping language models use a site's content at the moment someone asks, not at training.

## What it does and doesn't do

Google's guide says you don't need "AI text files, markup, or Markdown to appear in Google Search", and that Google Search itself doesn't use them. No major assistant has said its search ranks or cites pages by llms.txt.

Where it does help is with agents that have been pointed at your site to get something done: a developer's coding assistant reading your API docs, or an agent asked to work with your product. For those, a good llms.txt is a shortcut to the right pages. That's why software companies with documentation tend to publish one, and why Keats publishes [its own](https://getkeats.com/llms.txt), which lists every operation and workflow for agents that [run AEO through Keats](https://getkeats.com/guides/aeo-with-your-ai-agent).

## So what should you do?

- **Keep the structured data you have, and make it match.** Check product prices, availability, ratings and dates against the visible page.
- **Add Organization markup** if you don't have it, especially if your name is shared with another company.
- **Add an llms.txt if you have docs or a product agents will use.** Skip it if you're a [local business](https://getkeats.com/guides/aeo-for-local-business) or a [shop](https://getkeats.com/guides/aeo-for-ecommerce) with no API; it won't change what assistants say about you.
- **Spend the real effort on the page itself.** The answer near the top, specifics in text, current facts. Our guide to [writing content for AI search](https://getkeats.com/guides/answer-first-content) covers it.
- **Make sure crawlers can read you at all.** That's robots.txt, not markup: see [AI crawlers and robots.txt](https://getkeats.com/guides/ai-crawlers-robots-txt).

## How to tell if it mattered

If you add markup or an llms.txt and want to know whether it moved anything, treat it like any change: note the date, the questions it might affect, and the questions it can't, then compare over three to five weeks. Don't expect a verdict either way; a change too small to measure is the most likely result.


## Your agent does this

Ask your agent: "Check whether our product and pricing pages' structured data matches what the pages say, and list any mismatches." It reads the pages, compares the markup with the visible text, and drafts the fixes for review. Agents that work with Keats read its own llms.txt to find every operation.

## Sources

- [Google Search Central: Optimizing your website for generative AI features on Google Search](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide), read 2026-11-13
- [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features), read 2026-11-13
- [llmstxt.org: The /llms.txt file (Jeremy Howard, 2024)](https://llmstxt.org/), read 2026-11-13
- [Keats: llms.txt, the operations and workflows for agents](https://getkeats.com/llms.txt), read 2026-11-13