Do You Need Schema Markup or an llms.txt File to Show Up in AI Search?

If you've talked to a web designer or marketing company about AI search, you've probably heard two technical terms: schema markup and llms.txt. They're usually presented as special tricks that tell AI tools about your business.

Do they work? The honest answer, based on the best test we could find, is probably not the way people hope. This guide explains what each one is, what the evidence says, and what to do instead if your goal is getting named when customers ask AI tools for a contractor. It doesn't mean schema is useless. It means it isn't a shortcut.

What they are

Schema markup (structured data) is hidden code on a web page that labels facts: "this is a plumbing business, here is its phone number, here are its hours." Search engines use it for things like rich results. It's usually added in a format called JSON-LD.

llms.txt is a proposed plain-text file at the root of a website (yoursite.com/llms.txt) meant to give AI tools a summary of the site. It's a community idea, not an official standard from Google or OpenAI as far as we found.

What Google says

Google's page on its AI features (AI Overviews and AI Mode) is blunt:

• "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
• "You don't need to create new machine readable files, AI text files, or markup to appear in these features."
• "There's also no special schema.org structured data that you need to add."

It does say your page must be indexed and eligible to show in Search, and that any structured data you use should match the visible text of the page (Google).

What the best test found

Most claims about schema and AI come from correlations: sites cited by AI tend to have schema. That doesn't show schema caused the citations, because sites with good SEO tend to have both.

Ahrefs ran a more careful test. It identified 1,885 pages that added JSON-LD schema between August 2025 and March 2026, matched them against similar pages that didn't, and measured citations on Google AI Overviews, Google AI Mode and ChatGPT for 30 days before and after (Ahrefs). The results:

Ahrefs' conclusion: the data doesn't support adding schema mainly to win AI citations on pages that are already visible. Their caveats are fair: every page in the sample already had heavy AI Overview citations, so the test says nothing about pages that were never cited. Schema changes often came with other edits. And a separate experiment they cite found several AI systems ignored JSON-LD when fetching pages live.

A separate correlational analysis by Search Atlas (a vendor) also found no link between schema coverage and citation rates (Search Atlas).

What about llms.txt?

We found no test showing llms.txt affects citations, and Google's guidance above says no AI text files are needed. We found no statement from OpenAI that ChatGPT's search uses it. The Ahrefs article doesn't mention it. So treat it as unproven. If someone offers to add one, it's cheap, but don't pay much for it or expect results.

So is schema worthless?

No. It has real uses:

• Search features. Structured data can enable rich results in regular search, which is what it was designed for.
• Clean, consistent facts. Putting your name, address, phone, hours, services and service area in a standard format forces you to get them right.
• Low cost. A basic version takes a developer under an hour.

If you add it, keep it simple and honest. Here's the shape of a basic local business block (use the type that fits you, such as Plumber, HVACBusiness, RoofingContractor or Electrician):

{
  "@context": "https://schema.org",
  "@type": "Plumber",
  "name": "Your Exact Business Name",
  "url": "https://www.example.com",
  "telephone": "+1-555-555-0100",
  "address": { "@type": "PostalAddress", "streetAddress": "123 Main St",
    "addressLocality": "Garland", "addressRegion": "TX", "postalCode": "75040" },
  "areaServed": ["Garland", "Rowlett", "Sachse"],
  "openingHours": "Mo-Fr 08:00-17:00"
}

Whatever you put in must match what a visitor can see on the page and what's on your Google profile.

What to do instead

If your goal is being named when customers ask AI tools for help, these have better evidence behind them:

1. Let the crawler in. OpenAI says sites that block OAI-SearchBot won't appear in ChatGPT search. See how to allow it.
2. Be accurate on the sites AI cites. In our November 2026 study of 390 Texas markets, ChatGPT cited Angi in 78.5% and ContractorsRanked in 71.8% of markets. See the directories ChatGPT reads.
3. Keep reviews fresh and honest. See how many reviews and how recent.
4. Write clear service pages and FAQs, with real answers. See service pages and FAQs that answer "who should I call?".
5. Check results the right way. See the five-minute test.

A quick way to see whether your site has it

You don't need to read code. Ask your web person, "Does our site have LocalBusiness structured data, and does it match what's on the page?" Or paste your page address into Google's Rich Results Test, a free tool that shows whether structured data is present and valid. If it's there and accurate, leave it alone. If it's missing, adding a basic block is a small job, but put it after the higher-value fixes below, not before.

What to ask someone selling this

• "What evidence do you have that this changes ChatGPT's answers for a business like mine?"
• "Can you show a before and after, with the questions and dates?"
• "What else is in the package?"

If the pitch is mainly schema and llms.txt, be skeptical. See how to hire an AI-visibility agency.

FAQ

Will schema help my regular Google rankings? It can help search engines understand your page and may enable rich results. We aren't claiming a ranking boost.

My site builder adds schema automatically. Should I remove it? No. Just make sure it's accurate.

Is it worth paying for? Not as an AI strategy. As a small cleanup task, yes, if it's cheap.

See what's actually holding you back. Run a free scan at localpickai.com. It checks your website for AI-bot blocking and shows whether ChatGPT names you, who it names instead, which websites it read, and what to fix first. Free, no signup, about a minute.