I scanned 18 Indian diagnostic and pathology labs for AI visibility

I scanned the homepages of 18 Indian diagnostic and pathology lab websites, national chains, regional labs, and at-home-sample-collection players, using the same AI readability engine that powers my free Report Card tool. I set out to check 24 sites; 6 blocked automated access outright and were skipped rather than scored, so the 18 below are the honest sample, not a cherry-picked one. The median AI readability score came out to 79 out of 100, respectable on paper, but the range ran from 23 to 96, and the single biggest predictor of where a lab landed was whether it had structured data on the page at all: labs with it scored a median of 83, labs without it scored 46.

Who scored well, and who didn't

Of the 18 labs, 5 scored an A (85 or higher), 5 scored a B, 3 scored a C, 4 scored a D, and 1 scored an F. That is a genuinely split field: roughly a quarter of labs are doing this well, and roughly a quarter are barely readable to an AI engine at all.

The split gets sharper by lab type. National chains, the 9 labs in the sample with a multi-city or pan-India footprint, posted a median score of 82. At-home-sample-collection, app-first players (4 labs) came in at 74.5. Regional and local labs (5 labs) scored a median of just 49, the lowest of any group by a wide margin.

That last number is the one that should worry a regional lab owner. 'Blood test near me' or 'which lab does home collection in my area' is exactly the kind of question an AI assistant answers by leaning on structured, readable pages, and it is exactly the query type where a regional lab should have the advantage of proximity. Instead, the labs a nearby patient most needs to find are the ones an AI engine can read the least.

Structured data is the line that splits this sample

11 of the 18 labs have structured data (JSON-LD or schema markup) on their homepage; 7 do not. The labs with it scored a median of 83. The labs without it scored a median of 46. That is a 37-point gap, the widest single split in the whole dataset, and it comes down to one technical decision.

Structured data is a block of machine-readable text, usually JSON-LD, sitting in a page's code that tells a search or AI engine plainly what the page is: an Organization, a MedicalBusiness or DiagnosticLab, with its services, location, and sometimes price listed as explicit fields rather than implied by a page's layout. An AI engine crawling a page without it has to guess: parse headings, infer from images, hope the design conveys what a human eye reads instantly. A lab with clean structured data is handing the engine an answer key. A lab without it is asking the engine to solve a puzzle, and most of the time it just moves on to a site that didn't make it guess.

This is also the cheapest fix in the whole report. It does not require a redesign or new copy, just markup added to a page's existing code, and it is the single change most likely to move a lab from the 46 median to somewhere near the 83 median.

The basics are missing even on the better sites

Only 3 of the 18 labs have FAQPage schema, the markup that lets an AI engine pull a direct answer to a specific question straight from a page. 15 of 18 are missing it. And only 5 of 18 labs have a single, clear H1 heading, the most basic 'what is this page about' signal a page can send; 13 of 18 fail it. This is not a niche gap. A missing or duplicated H1 is the kind of thing that shows up even on sites that otherwise look polished and score a B.

The questions patients actually ask are predictable: what does a thyroid test cost, does this lab do home sample collection, is it NABL accredited, how long does a report take, how do I access my report once it's ready. FAQPage schema exists specifically to let an AI engine answer exactly these questions by quoting a lab's own page. Right now, almost nobody in this sample has wired it up, which means the questions patients are actually typing into ChatGPT are the ones the labs are least prepared to answer.

The smaller gaps in the sample:

• Open Graph tags on only 11 of 18 labs, so links shared on social or WhatsApp render with no preview image or title.
• A clear, unique page title on just 10 of 18 sites.
• FAQ markup on only 3 of 18 labs, leaving the most-asked patient questions unanswerable by AI.
• One clean H1 heading on only 5 of 18 homepages, the weakest fundamental in the sample.
• 6 of the 24 labs I attempted to scan blocked automated access entirely, meaning even a basic AI crawler may be locked out before it reads a single word.

What to fix, for the sites still behind

1. Add JSON-LD structured data for Organization and MedicalBusiness or DiagnosticLab, with your services and prices written as text fields, not buried in an image. The schema generator builds this block for you in a minute.
2. Give every page one single, clear H1 heading that states plainly what the page is.
3. Add FAQPage schema answering the real questions patients type into AI assistants: test prices, home sample collection, NABL accreditation, turnaround time, and how to access a report.
4. Write test prices and turnaround times as plain text on the page itself, not inside an image or PDF that an AI engine can't read.
5. Add Open Graph tags, then run the free Report Card to confirm the fix actually shows up in a scan.

Method and limits

A few honest limits. This scan covers only the homepage of each lab, a single snapshot in time, not the full site or how it changes week to week. It runs on my open-source analyzePage engine, the same one behind the free Report Card tool, and it measures machine-readability: whether an AI engine can parse structure, headings, and schema on a page. It says nothing about a lab's medical accuracy, service quality, or accreditation itself, only whether that information is written in a form an AI engine can read. I attempted to scan 24 labs; 6 blocked automated access and were excluded rather than scored, leaving the 18 reported here. No individual lab is named in this piece; the numbers are reported in aggregate.

Frequently asked questions