This post is for diagnostic chain owners, pathology lab heads and the marketing and web teams who maintain their test catalogues. It covers a narrow problem: when someone asks an AI system "how much is a thyroid profile in Delhi", "do I need to fast for a lipid profile" or "when will my report be ready", which lab's page gets quoted, and how a lab can make its test pages safe and useful to quote.
Lab test queries are among the most frequent health searches. They are also unusually exposed to error. A price may be out of date, a fasting instruction may be oversimplified, and a reference range copied without context may reassure someone who should have called a doctor. That last problem is not hypothetical, as the liver-test example below shows. A lab page that handles these details carefully is more useful to patients and, we think, more likely to be used by systems that are trying to avoid repeating bad information.
We describe a method and a page template. We do not report results from testing, because answers differ by surface, location and day. Run the test described here on your own catalogue to get your own baseline.
What happened with liver-test ranges
In January 2026 Google removed AI Overviews for queries such as "what is the normal range for liver blood tests" after a Guardian investigation. As reported by Storyboard18, the Guardian's concern was that the summaries gave numerical ranges without accounting for factors such as age, sex or ethnicity, which could lead someone to assume their results were normal. Google said its clinicians had reviewed the flagged queries and often found the information was not inaccurate and was supported by high-quality sites. Variations of the query reportedly still triggered summaries for a time, and the same questions were still offered in AI Mode.
The lesson for a lab is not that reference ranges are forbidden. It is that a bare range on a public page, without the context that makes it interpretable, is a liability. A lab's own report ranges depend on its method, instrument and population. A page that publishes a generic range may contradict the patient's own report, or encourage self-interpretation. Treat this as a clinical governance matter and have your pathologist decide what appears on the public page.
How AI answers handle the three common lab queries
The three families of queries below behave differently, and each needs a different page element.
Price queries ("CBC test price", "HbA1c cost near me"). These are comparison queries. The system looks for a figure, a location and often a provider. Prices change, vary by branch and by home collection, and may differ between a single test and a package. If your page shows no price, or an unlabelled one, an aggregator's number will be used instead.
Preparation queries ("fasting for lipid profile", "can I drink water before blood test"). These are safety-adjacent. The answer must be accurate, specific to the test and cautious about medication and conditions. Generic advice from a forum or a video will often fill the gap if your page does not.
Turnaround queries ("TAT for thyroid profile", "when will my report come"). The answer depends on test, batch time, sample cut-off and location. A vague "same day" invites a complaint.
Step 1: Test 50 queries across AI surfaces
The method is the same as for any AI search review, with lab-specific groupings. Build 50 queries across your top 25 tests, two per test: one price-intent and one preparation or turnaround query. Add city and area variants for your main markets, and a few in Hindi or Hinglish if your patients use them.
Run each on Google (note whether an AI Overview or AI Mode answer appears), and on one or two chat assistants. Use a clean browser, set the location to a city where you operate and record the date.
For each result, log:
- Whether an AI answer appeared.
- The price stated, if any, and whether it matches your current list.
- The preparation instruction stated, and whether your pathologist would sign it off.
- The turnaround stated.
- Which sources were cited, grouped as your site, competing lab, aggregator, health publisher, government or other.
- Whether any range or interpretation was given, and whether it carried context.
The error log is the most valuable part. Every wrong price, wrong fasting instruction or unqualified range that your competitors or aggregators supply is a content gap you can close with a clear, reviewed page. Where the answer is wrong about your own lab, you cannot edit it, but you can publish a dated, reviewed page that makes the correct information easier to quote.
Google's AI features documentation says that to appear as a supporting link, a page must be indexed and eligible for a snippet, and that no extra markup or special files are needed. The work, therefore, is in the content and its structure.
Step 2: A test-page template that holds up
Use one page per test or panel, with the same structure across the catalogue. A template makes review easier and gives every page the same labelled sections.
Header block. Test name, common aliases (for example, "HbA1c", "glycosylated haemoglobin") and the department. Aliases matter because patients and reports use different names.
What the test is for. Two or three plain sentences on what the test measures and why a doctor might order it. This is clinical content and needs pathologist review.
Price block. Price in rupees, what it covers, whether home collection adds a charge, the branch or city it applies to, and the date it was last updated. If prices differ by branch, show them in a table or state clearly that the page shows the price for a named city.
Preparation block. Fasting duration if required, whether water is allowed, medication and supplement notes marked "ask your doctor before changing medicines", timing of sample collection where it matters (some tests are time-sensitive), and who should call before booking, such as pregnant patients or those on specific treatment.
Sample and collection. Sample type, whether home collection is available, and minimum notice.
Turnaround block. State what the time means: "reported by 6 pm for samples received before 11 am" is a statement a patient can check. "24 hours" is not.
Results block. What the report will show, and a plain statement that interpretation depends on age, sex, health history and the lab's method, and should be done with a doctor. If you publish ranges at all, publish the lab's own, with the method, and with the age and sex bands, after pathologist sign-off. Otherwise, leave them out and say why.
Booking and contact. One clear action, with the correct branch.
Review line. "Clinically reviewed by [named pathologist, qualification]. Last updated [date]." It must be true. A named reviewer who has never seen the page damages trust.
