
AI Search Readiness: What I Found Auditing 53 Local Healthcare Websites
Last September I ran an AI search readiness check on 53 local healthcare websites: chiropractors, physical therapy clinics, dentists and wellness practices, mostly in Chicago’s western suburbs. I wanted to know one thing. When ChatGPT, Gemini or a Google AI Overview tries to work out who a practice is and what it does, how much help does the website give it?
The short answer: not much. The sites were technically sound. Every one loaded over HTTPS and every one was mobile-friendly. But the parts that tell a machine “this is a chiropractor at this address, with these profiles, answering these questions” were missing far more often than they were present. Here is what I found, with the real numbers.

How I ran the AI search readiness check
Each homepage went through MRK Audit, the audit tool I built, between September 14 and 21, 2026. It runs about 40 automated checks covering technical SEO, structured data, content structure, AI crawler access and answer-readiness, and it gives three scores out of 100: overall, GEO (generative engine readiness) and AEO (answer engine readiness).
The sample: 19 chiropractors, 14 physical therapy clinics, 9 dental practices and 11 other health and wellness practices (integrative clinics, a med spa, wellness centers). I reviewed these sites while researching local practices, so they were not picked at random, and I checked homepages only. Treat the numbers as a snapshot of 53 real sites, not a national benchmark. I’m not naming any practice.
The headline: technically fine, invisible as an entity
The median technical score was 81%. The median GEO score was 38 and the median AEO score was 35. Search engines can crawl these sites. They just can’t learn much about the business from them.
| What I checked | Sites missing it |
|---|---|
| Any structured data (JSON-LD) at all | 14 of 53 (26%) |
| A business-type entity (LocalBusiness, Dentist, Chiropractor, MedicalBusiness and similar) | 19 of 53 (36%) |
| sameAs links tying the site to its profiles | 31 of 53 (58%) |
| Review or rating markup | 48 of 53 (91%) |
| Question-style headings | 17 of 53 (32%) |
| A table anywhere on the page | 51 of 53 (96%) |
| Images with alt text (fewer than 80% covered) | 16 of 53 (30%) |
| Exactly one H1 | 14 of 53 (26%) got this wrong |
| Secure resources only (no http files on an https page) | 12 of 53 (23%) loaded http files |
Only 20 of the 53 sites (38%) had the full basic set: a business entity, links to their profiles, and a phone number and address in the markup. That set is how a practice tells Google, and every system that reads Google’s index, exactly who it is.
Chiropractors had the weakest entity data
The gap varied a lot by profession:
- Chiropractors: 9 of 19 (47%) had no business-type entity. Median GEO score 38.
- Physical therapy clinics: 5 of 14 (36%) had no business-type entity. Median GEO score 40.5.
- Dental practices: 1 of 9 (11%) had no business-type entity. Median GEO score 54.
My read, which this sample can’t prove: several of the dental sites were built on dental-specific website platforms, which tend to include Dentist markup. More of the chiropractic and PT sites looked like general templates that leave it out. If you run a chiropractic office, the fix covered on my chiropractor SEO services page is usually the first thing to do. For clinics, my physical therapy SEO page covers the same ground.
AI crawlers were rarely blocked
This one surprised me. A lot of AI search advice starts with “check your robots.txt”. In this sample only 2 of 53 sites (4%) blocked any AI crawler. Both blocked several at once (GPTBot, ClaudeBot, Google-Extended, Common Crawl), which looks like a host or plugin default rather than a deliberate choice. For most practices, crawler access is not the problem. Being understood once the crawler arrives is.
The content gap: few direct answers
Patients ask AI tools full questions: “do I need a referral for physical therapy in Illinois”, “how much does a crown cost without insurance”, “is a chiropractor covered by Medicare”. A homepage that never phrases anything as a question and never answers one plainly gives an answer engine nothing to quote.
Almost a third of the sites had no question-style headings at all, and 15 of the 53 homepages had fewer than 600 words. Only 5 had FAQ markup, but markup is not what matters most here. What matters is whether the page answers the questions patients actually ask, in plain sentences.
What works: five fixes from the data
1. Add the right business entity
Use the most specific type that fits (Dentist, Chiropractor, Physician or MedicalBusiness), with name, address, phone, hours and URL that match your Google Business Profile exactly. Google’s local business structured data guide lists the properties it reads.
2. Connect your profiles with sameAs
Link the entity to your Google Business Profile, Healthgrades or Zocdoc listing, Facebook and LinkedIn. This is how a machine confirms that the website, the map listing and the review profiles are one practice. Most sites in this sample skipped it.
3. Answer five patient questions in plain words
Pick the questions your front desk hears every week and answer each in two or three sentences on the page where it belongs. That gives answer engines something accurate to quote, which is the core of answer engine optimization.
4. Fix the small technical leaks
One H1, alt text on every meaningful image, no http files on an https page. A quarter of the sites had at least one of these issues. Each takes minutes to fix.
5. Name the people
Healthcare is a trust category. List your practitioners with credentials on the site and mark them up as people connected to the practice. A clinic with named, verifiable clinicians gives AI systems a reason to trust and cite it, which is what generative engine optimization is about.
How long the fixes take to show up
The markup changes take a day or two to make. Google usually re-crawls a small practice’s homepage within one to three weeks, and you can speed that up by requesting indexing in Search Console. Changes in AI answers lag behind that, often by a month or more, because those systems also lean on reviews and mentions elsewhere. I’d judge the work after about 60 days, not 6.
What it costs
If you do it yourself, the entity markup and the question-and-answer content cost nothing but time: roughly a day for most practice sites. If you hire it out, it fits inside a normal small business SEO engagement, which on my side runs $500 to $2,000 a month depending on the number of locations and pages.
Do it yourself or hire someone?
If your site runs on WordPress with an SEO plugin, you can add most of the entity data yourself through the plugin’s local business settings. Hire help if your site has several locations, if your markup already exists but contradicts your Business Profile, or if you’d rather spend the day with patients. Contradictory data is worse than none, and it’s the most common problem I fix. My dental SEO services page shows how I handle it for practices.
Start this week
- Run your homepage through Google’s Rich Results Test and check whether a business entity shows up.
- Compare the name, address, phone and hours in it with your Google Business Profile, character for character.
- Write down the five questions patients ask most, and answer them on your site.
- Check your robots.txt once, just to rule out the rare block.
The practices that get named in AI answers over the next year won’t necessarily be the biggest. They’ll be the ones whose websites state clearly what they are, where they are and what they know.
Frequently asked questions
What is AI search readiness?
AI search readiness is how easily systems like Google AI Overviews, ChatGPT and Gemini can understand who a business is and quote its pages. It depends on clear business entity data, links to the business’s profiles, plain answers to common questions and normal technical health, not on any special AI file.
How many of the 53 healthcare websites were AI search ready?
Only 20 of the 53 sites (38%) had the basic set: a business-type entity, sameAs links to their profiles, and a phone number and address in their structured data. Median scores were 38 out of 100 for generative engine readiness and 35 for answer engine readiness, even though the median technical score was 81%.
Do I need to unblock AI crawlers in robots.txt?
Check once, but it is rarely the problem. Only 2 of the 53 sites in this check blocked any AI crawler. Both blocked several at once, which usually points to a host or plugin default. Most practices are crawlable; what they lack is data that explains the business.
Which schema type should a chiropractor, dentist or physical therapist use?
Use the most specific type that fits: Dentist for dental practices, Chiropractor for chiropractic offices, and MedicalBusiness or a medical clinic type for physical therapy. Include name, address, phone, hours and URL exactly as they appear on your Google Business Profile, plus sameAs links to your profiles.
How long until AI answers start mentioning my practice?
Google usually re-crawls a small practice homepage within a few weeks, but AI answers lag behind because they also rely on reviews and mentions on other sites. Judge the work after about 60 days, and keep earning genuine reviews and local mentions alongside the on-site fixes.
Free SEO audit · No signup
Reading about SEO is good. Knowing your score is better.
Run a free SEO + GEO + AEO audit of your website — results in about 15 seconds, downloadable as a PDF.