Proven Advice vs Hype, and Reading Site Readiness Correctly

In this lesson

AI visibility is a young field. Every week brings a new file, tag or "AI-only" setting someone says you need. This lesson gives you a way to decide whether a recommendation deserves your time. Then it covers how to read site readiness correctly.

A three-question filter

When you come across a recommendation, ask:

1. Is it in the provider's own documentation? Do OpenAI, Anthropic, Google or Perplexity describe this setting in their docs? A blog post or one tool's checklist doesn't count.
2. Has its effect been shown? Is there a measurement with a disclosed method showing it changes answers? "It makes sense" isn't evidence.
3. Is something more basic missing? If a search crawler is blocked in robots.txt, or your pricing doesn't come from the server, fix that before you spend time on anything new.

If the answer to the first two is "no", the recommendation may be harmless, but it's low priority.

Example: llms.txt

What is it? llms.txt is a community proposal for a text file at the root of a site, meant to give AI systems a summary of it. It isn't an official standard.

Where does it stand (as of October 2026)?

• Google says AI features don't need special AI text files. Source: Google, AI features and your website.
• It hasn't been shown that major AI engines rely on the file for their answers.

Through the filter:

1. Provider documentation: no support from the major engines, and Google says its AI features don't need such a file.
2. Shown effect: none.
3. More basic gaps: on most sites, robots.txt, server-readable content and consistent brand facts come first.

Verdict: low priority. Adding the file does no known harm, but don't spend time on it until the basics above are done. If an audit checklist flags a missing llms.txt, read it as information, not as a problem to fix.

The same filter works for other ideas, such as "AI meta tags", separate pages served only to AI crawlers, or scores a single tool made up. Showing crawlers different content from what users see is also risky. Don't do it.

Site readiness is not SOV

Everything in this module falls under site readiness: crawler access, server-readable content, structured data. These are the conditions your site needs to be readable.

SOV measures something else: how often you're mentioned relative to your confirmed competitors in answers to non-branded questions. You saw this in Module 2.

The two answer different questions:

• Site readiness: "Can my site be read?"
• SOV: "How much of the answers do I get compared with my competitors?"

Readiness opens a door. It doesn't measure who walks through it or what's said inside.

Two representative cases

Acme Helpdesk and Northwind Boards are fictional products. The cases are examples.

• Acme Helpdesk passes every check in this module: deliberate robots.txt, server-delivered content, accurate structured data. Yet it's rarely mentioned in answers to non-branded questions. Module 3 explains why: few third-party sources talk about it.
• Northwind Boards has a few gaps on its site, yet it's mentioned often, thanks to review platforms and Reddit threads.

In neither case does site readiness explain the outcome on its own. That's why, after readiness work, visibility has to be measured separately.

Reading a readiness checklist

A site readiness checklist, including the Site Readiness section of SOV Tracker's free scorecard, shows check results, not a score: each check is reported as passed, issue or info, and checks that couldn't be measured aren't counted as issues. Site readiness isn't SOV, isn't folded into any of the seven core AI metrics, and doesn't guarantee that AI engines will mention your brand. Definition: AI Site Readiness.

When you read any checklist, rank the issues. They don't all carry the same weight.

Order of priority

1. Access. No search crawler is blocked by accident. robots.txt doesn't return 5xx errors. Bot protection doesn't block crawlers you want.
2. Readable key facts. Prices, plan limits and policies are in the server-delivered HTML as text.
3. Structured data and consistency. Core facts are marked up and match the page and your brand fact sheet.
4. Weakly supported ideas. Things like llms.txt come last, and only if time allows.

Summary

• Filter any recommendation with three questions: is it in provider docs, has its effect been shown, is something more basic missing?
• llms.txt is a community proposal. Google says its AI features don't need such a file, and it hasn't been shown that major engines rely on it. It's low priority.
• Don't show crawlers different content from what users see.
• Site readiness answers "Can my site be read?". SOV answers "How often am I mentioned compared with my competitors?"
• Site readiness is a checklist. It isn't SOV and doesn't guarantee visibility.
• Order: access, readable key facts, structured data and consistency, and weakly supported ideas last.