Teardown · Lab networks
2,302 lab pages. Every one named “Detail”.
Can AI find the right lab? One network wrote a note for the AI engines. The other tells Google its labs are all the same page.
Research by ClinAds · 7 min read · September 28, 2026
Two national lab networks, and one patient search: blood test lab near me open saturday. Labcorp keeps its patient service centers on locations.labcorp.com. Quest Diagnostics runs 2,302 lab pages on questdiagnostics.com. Either should be easy for an AI assistant to name.
Only one of them is. Labcorp wrote a note for the AI engines and gives every lab its own title, its own address and hours a machine can read. Every Quest lab page is called “Detail”, with the same description and the same canonical URL as 2,301 others.
Five questions: Labcorp answers four, Quest answers two
We scored the page a patient actually lands on, read before JavaScript runs.
Labcorp locations.labcorp.com
4/5
- ✓Where is it?
- ✕Are the hours in plain text?
- ✓Can a machine read it as a lab?
- ✓Does it have its own name?
- ✓Does it point to itself?
Quest questdiagnostics.com
2/5
- ✓Where is it?
- ✓Are the hours in plain text?
- ✕Can a machine read it as a lab?
- ✕Does it have its own name?
- ✕Does it point to itself?
Method: 40 location pages sampled at random from each network, scored on the raw page before JavaScript runs.
Labcorp gives every lab a name and a listing of its own
Each Labcorp lab page carries a business listing a machine can read: what it is, where it stands, when it opens. The title names the town and the ZIP, so the search result reads like a place instead of a database row.
A Labcorp lab page · title and page data
<title>Laboratory Testing in Novato 94945 | Labcorp</title>
"@type": "MedicalBusiness",
"openingHoursSpecification": [ … ], "geo": { … }
And Labcorp wrote a file addressed to the AI engines
Almost no healthcare brand publishes one. Labcorp’s llms.txt says what the company is and points assistants at the pages that matter, including the one that finds a lab.
labcorp.com/llms.txt · 69 lines
# Labcorp (Laboratory Corporation of America)
This file provides a structured index of official Labcorp
endpoints for use by Large Language Models, AI search agents…
## Patients & Individuals
- [Labs & Appointments](…): locate patient service centers
Every Quest lab page is called “Detail”
The title, the description and the heading are the same word on all of them. In a search result, 2,302 labs look like one page repeated.
Then each Quest lab page tells Google it is a copy of one other page
A canonical tag is how a page says “this is the original”. Every Quest location page points at the same URL instead of at itself, which asks search engines to treat 2,302 labs as duplicates of a single page.
The canonical tag on every Quest lab page we checked
/locations/detail.html/COH/02025/75/1 →
<link rel="canonical" href="/locations/detail">
/locations/detail.html/JV/32210/75/1 →
<link rel="canonical" href="/locations/detail">What that costs: the address, the hours and the phone number are all on the page in plain words. The page then asks not to be counted as its own.
The scorecard: seven checks to three, and one failure they share
Nine checks, the same method, both networks.
| Check | Labcorp | Quest |
|---|---|---|
| A note written for AI enginesAn llms.txt on the domain | ✓ | ✕ |
| A business listing in the page dataWhat a machine reads to place a lab | ✓ | ✕ |
| Hours a machine can parseNot only printed for a reader | ✓ | ✕ |
| Hours printed for a readerVisible words, not only code | ✕ | ✓ |
| Its own title, one per labHow the result reads in search | ✓ | ✕ |
| Points to itself as the originalThe canonical tag every page needs | ✓ | ✕ |
| Location pages in a sitemapSo a crawler knows they exist | ✓ | ✓ |
| robots.txt serves real rulesNot an error, not a blank | ✓ | ✓ |
| Dead location URLs return an errorNot “200 OK” for a lab that isn’t there | ✕ | ✕ |
Why this matters to a clinic
A page that won’t say its own name can’t be recommended. Quest has the facts a patient needs on every page, and spends them on a title that says nothing and a tag that points search engines elsewhere. Labcorp wrote the same facts in a form a machine can lift.
The gap is not budget: a title, a canonical tag and a listing block is one page of work.
- We search for your clinic the way a patient would. Your city and your competitors, on Google and on the AI assistants.
- You get the gaps in plain sentences. One list you can hand to whoever runs your website.
- It takes a day. No commitment, no cost.
Download the slides (PDF, 620 KB)
Want a free AI visibility audit for your practice?
Free, and it takes about a day. We check your city and your competitors the way a patient does, then send back every gap in plain sentences.
Get my free auditClincy completes new-patient intake by voice or SMS before the visit, then writes it back to your PMS.
Book CallMore research
Teardown · Orthopedic care
2,054 clinicians. None marked as people.
Hanger publishes 2,054 certified clinicians and marks none as people. The arch support shop down the road explains itself in plainer words.
Teardown · Eye care
986 store pages. 56 characters each.
Both chains file the listing machines read. LensCrafters names the vision plans on every store; a MyEyeDr. store page is one line long.
Teardown · Dental groups
One has no front door. The other broke the lock.
Heartland’s website is written for dentists, not patients. Pacific Dental built the directory, then broke every listing with one stray character.
Teardown · Physical therapy
Same size. Opposite blind spots.
ATI hands machines the address, hours and map pin but names no insurer. Athletico names the plans and ships no machine-readable data.
Teardown · Urgent care
Ask AI for urgent care. Neither chain has the full answer.
CityMD’s pages answer four of five patient questions, MinuteClinic’s answer three. Each leaves out what the other gets right.
Teardown · Fort Worth, TX
Aspen Dental is losing Fort Worth to Brident
Brident’s page answers all five things a patient asks. Aspen’s answers none, across 71 Texas cities.