In short
We checked 85 live SEO agency websites for the signals that make a site readable by AI engines. The basics are everywhere: 87 percent have structured data, 76 percent have an Organization entity, and all of them use HTTPS. The depth is not: only 15 percent use FAQ schema (the format AI engines most readily quote), and where an llms.txt file exists (about 40 percent), it is almost always auto-generated by an SEO plugin rather than written on purpose.
How we ran it
We sampled 90 SEO and digital-marketing agencies from real searches across twelve cities in one European market (Romania), 85 of which were reachable. For each one we checked, automatically and without naming any agency, six objective signals: structured data (JSON-LD), an Organization-type entity, sameAs links to official profiles, FAQ schema, an llms.txt file, and HTTPS. This measures the state of the market, it does not judge any single agency.
What we found, across 85 reachable sites
- Structured data (JSON-LD): 87 percent (74 of 85).
- An Organization-type entity: 76 percent (65).
- sameAs links to official profiles: 60 percent (51).
- FAQ schema: 15 percent (13). This is the biggest gap, and it matters, because FAQ content is one of the formats AI engines most often lift as an answer.
- An llms.txt file: about 40 percent (34). Surprisingly high, and the reason is the interesting part.
- HTTPS: 100 percent (85). Basic security is no longer a differentiator for anyone.
The surprise: plugins are shipping llms.txt for you
We expected the llms.txt file, the one that tells AI engines who you are, to be rare. It was on about 40 percent of sites. When we opened them, almost all were auto-generated by an SEO plugin: Yoast, Rank Math, or All in One SEO. Many began literally with a line like "Generated by Yoast SEO" or "Generated by Rank Math."
In other words, the popular SEO plugins have quietly started placing an llms.txt on every site that runs them. It is a good default. But the generated file is generic, usually a list of posts, not a clear statement of what the business actually is.
What it means
The takeaway is not that these agencies are behind. The basics are there. The takeaway is that the floor is rising on its own, through plugins, so the advantage is moving to depth. Structured data no longer sets you apart, almost everyone has it. What sets you apart now is what the plugins do not do for you:
- a clear, consistent entity with real links to your actual profiles, not a generic block;
- FAQ schema, used by only 15 percent, in the format AI quotes;
- an llms.txt written on purpose that states what you are, not an auto-generated list of posts;
- content that answers the real questions people ask, in their own words.
Where most stop at the plugin defaults, the difference is the work you do beyond them.
Frequently asked questions
What is an llms.txt file?
It is a plain-text file that tells AI engines who you are, what you do, and where the important content is. The idea is useful, but its value depends on how clearly it is written, not just on whether it exists.
Why does FAQ schema matter for AI?
AI engines often lift answers from clearly marked question-and-answer content. Marking up a real FAQ section increases the chance your answer is the one shown.
Does having structured data guarantee an AI will cite me?
No. It is necessary but not sufficient, most sites already have it. The difference comes from depth, a clear entity, and content that answers real questions.
How were the agencies chosen?
From real searches across twelve cities, with no agency named. The study describes the general state of the market rather than judging individual companies.