What 16 clinic websites revealed about AI visibility
I scanned 35 Indian clinic sites for how readable they are to AI answer engines and published the 16 that scored 85 or below. This is what those 16 rows say when you add them up. Every number here is recomputed from the public index at build time, so it cannot drift from the data.
The AI Visibility Index lists each clinic with its score and the method attached. A single row is a diagnosis. Sixteen rows together are a pattern, and the pattern is more useful than any one site: it says which gaps are common, which are rare, and what separates the sites that read well from the ones that do not.
The short version, before the charts:
- Median score is 77 out of 100, across a range of 42 to 84.
- None of the 16 reached an A. The best site scored 84.
- 25% publish no structured data, and those sites cluster at the bottom of the range.
- Structured data splits the set: median 54 without it, 80.5 with it.
The headline numbers
The average site here is readable but leaky: a median of 77 means the typical clinic is missing roughly a quarter of the signals an answer engine looks for. And not one of the 16 reached an A. The ceiling in this group was 84, which is a good B, not a great score. These are not broken websites. They are legible-enough websites that leave money on the table every time someone asks an AI for a recommendation.
How the scores spread
Clinics by score band
The mass sits in the 70s and 80s: competent sites with a missing signal or two. The tail below 60 is small but sharp, and as the next chart shows, it is not random which sites fall there.
Structured data is the dividing line
Median score, by whether the site publishes structured data
The 4 clinics that publish no structured data have a median of 54. The 12 that do have a median of 80.5, a gap of 26 points. Structured data is not the only thing that matters, but in this sample it is the clearest single fault line: a block of JSON-LD in the page head, naming the practice and its services, is the highest-leverage fix on almost every low-scoring report.
Grades, in full
Clinics by grade band
By speciality
Median score, by clinic type
Cosmetic sites read best in this group and Hair Transplant sites worst, but every band is small enough that one site moves the number. The speciality matters less than the build: a Hair Transplant site with clean structured data still beats a Cosmetic site without it.
Method
Each site was fetched once and its homepage HTML checked against a fixed list of signals: structured data (JSON-LD), recognised schema types, FAQ markup, a title in range, Open Graph tags, a single clear H1, heading structure, semantic HTML, readable in-HTML text, and whether AI crawlers are allowed. Each signal is worth a fixed number of points; the total is normalised to 100. The check runs no JavaScript, deliberately, because the major AI crawlers do not either. The scan is deterministic: the same page scores the same every time, and you can re-run any row yourself from the index.
Limitations
This is a sample of 16, and a deliberately skewed one: it is the sites that scored 85 or below, so the median here is lower than it would be across all 35 scanned. It reads the homepage only, not the whole site. It measures machine-readability, not care, price, or outcomes: a clinic can be excellent and still score a D, which is exactly the problem, because an answer engine cannot tell the difference and picks the legible one. And the per-speciality and per-city cuts are tiny. Treat the structured-data finding as the robust one and the rest as texture.
Frequently asked questions
How many clinics does this cover?
16. They are the clinic websites from a scan of 35 that scored 85 or below, the published rows of the AI Visibility Index. The 19 that scored above 85 are not listed, so this is a study of the sites with room to improve, not a random sample.
What does the score measure?
How readable a homepage is to AI answer engines like ChatGPT and Perplexity: structured data, Open Graph tags, a clear title and headings, FAQ markup, and whether the important text is in the HTML rather than behind JavaScript. It is not a measure of the clinic, only of the website's machine-readability.
Why did no clinic score an A?
An A needs 85 or more. The best in this set scored 84. Every site was missing at least one high-value signal, most often structured data or FAQ markup. That is the point of publishing this: the gap is common and fixable.
Is structured data really the dividing line?
In this sample, yes. The 4 clinics that publish no structured data have a median of 54, against 80.5 for the 12 that do. Structured data is the single machine-readable statement of what a business is, and its absence tracks the lowest scores here.
Can I check my own clinic?
Yes, free and with no signup. Paste your URL into the Report Card and you get the same 0 to 100 score and a ranked fix list in about ten seconds. Nothing is stored.