We measured twelve fertility clinics. The three with llms.txt are the three you have heard of.
A couple at 2 a.m. after a failed cycle asks an assistant: “which IVF clinic has the best success rates for a 38-year-old?” The machine names the clinics it can read. In September 2026 we checked which of twelve big clinics a machine can actually read — and the pattern is the opposite of what you would expect from a sector built on trust.
The 2 a.m. search
Nobody researches IVF during business hours. The searches happen late, after a negative test or a cancelled cycle, when a couple is too tired to be patient and too scared to click the wrong thing. They ask the phone the question they cannot ask their families. The clinic that has already answered — in words a machine can lift and a human can trust — is the one they remember at 9 a.m.
What we measured
Same method as always, dated and public: robots.txt, llms.txt and the homepage of twelve clinics, September 8, 2026, single vantage point, honest user-agent. Raw data: per-clinic CSV · full JSON.
Zero of twelve wrote an AI policy
Not one clinic names a single AI crawler in robots.txt. 12 of 12 are open by default — which is not a decision, it is an absence of one. For a sector whose entire product is trust and evidence, that is a strange thing to leave to default settings. The machines can read everything; nobody has chosen what they should read.
The three names with llms.txt are the three biggest
Shady Grove Fertility, CCRM and ORM — arguably the three most recognized fertility brands in the United States — are the 3 clinics serving hand-written llms.txt. This is the finding that stays with you: the market leaders are not just better marketed, they are more machine-readable. Whether that is cause or effect, the gap between the famous and the findable is now measurable.
RMA renders 34 words without JavaScript
RMA Network’s homepage — a major national brand — produces 34 words once scripts are stripped. Thirty-four. A machine cannot summarize what it cannot read, and a patient cannot skim what never renders. The irony is sharp: some of the most technically advanced clinics run the least readable websites, because the site was built to impress a human in a demo, not to be read by the thing a patient actually asks.
What this means for a clinic
Patients are already asking machines. The clinics that answer — plainly, honestly, with dated evidence — will be the names those machines give. The work is not complicated: success rates presented with their real context, cost pages that answer the question, and a site that loads its words without JavaScript. The clinics that do this first in their city own the 2 a.m. search for years.
How did you measure the clinics?
We fetched robots.txt, llms.txt and each homepage on September 8, 2026 from a single vantage point, and recorded AI policy, llms.txt type and no-JavaScript readability. Raw data is public.
Why is no-JavaScript readability important?
Because it is our proxy for what a machine can read. A homepage that renders to a few words without scripts is invisible to the crawlers and assistants that answer patient questions.
Does llms.txt guarantee patients find a clinic?
No engine is required to use it. It is a signal of a deliberate decision to be readable — and the clinics that made that decision are, in our sample, the market leaders.
Can you guarantee a clinic gets cited by AI?
No honest agency can. We build the readable, structured, verifiable signals that make it likely, and we measure the result.
Want me to look at your website with this lens?
Send the site through the free audit first. If the business is serious and the opportunity is real, Manas can review what is helping, what is missing and what should be fixed first.
Related SynapseIN resources
Every number comes from our September 8, 2026 crawl. Raw data: research/fertility-clinic-ai-visibility-2026.csv. Corrections are dated and additive.