Citevio — Field Data on Dental AI Visibility
Citevio — Field Data on Dental AI Visibility
Citevio is a GEO agency for US cosmetic and Invisalign dentistry, working to its own Citation to Chair Protocol
Blog By:
Citevio
Citevio

Schema markup will not buy you a citation. It still has a job.

8/18/2026 8:55:00 PM   |   Comments: 0   |   Views: 36
Short answer: Across 527 US dental practice websites in seven metros plus nine smaller nearby towns, scanned between 6 and 26 June 2026, 284 sites — 53.9% — carried no LocalBusiness or Dentist schema at all, and only 18 sites, 3.4%, had the complete configuration we measured. We did not test whether adding it changes AI visibility, and anyone telling you it does is selling past their evidence.

Structured data is a machine-readable label. It can state what a business is and attach facts to a consistent identity. It cannot compel an assistant to cite, recommend, or rank that business, and one published experiment suggests the connection is looser than the marketing around it implies.

What the audit found

Measured schema staten of 527%
No LocalBusiness or Dentist schema present28453.9
Present but incomplete22542.7
Complete measured configuration183.4
The denominator: 527 US dental practice websites across seven metros plus nine smaller nearby towns, scanned 6–26 June 2026. The sample began at 611 records; 84 were dropped for carrying no dental signal, which is the kind of cleaning step that belongs in public rather than in a footnote nobody publishes.

The middle row is the one usually lost. "Does the site have schema?" is asked as a yes-or-no question, and the answer here has three states, not two. Slightly more than four sites in ten had something in place that our check scored as incomplete. A plugin was probably switched on. A field, a type, or a relationship was missing. On a binary audit those 225 sites report back as a tick, and the practice believes a job is finished that is roughly half done.

Only 18 sites out of 527 — 3.4% — passed the complete check. That is a low number, and it is worth being careful about what makes it low: it reflects the configuration we chose to measure, on the day we measured it. A different checklist would produce a different figure.

The layer next door was mostly fine

The same 527 sites were also checked for their robots.txt state, and the contrast is instructive. There, 492 sites — 93.4% — came back OK, with 20 at WARN (3.8%) and 15 at FAIL (2.8%).

So on the layer everyone has been told to check for a decade, the sample is in good shape. On the labelling layer, more than half the sample has nothing at all. The industry's attention and the sample's actual gaps are pointing in different directions — though note that this is a comparison of two rates in one sample, not evidence that one of them matters more to any engine.

What schema actually does

It removes ambiguity. A page can say "we've served families here for thirty years" in prose without ever declaring the business type, the legal name, the address, or how one page relates to another. Structured data hands a machine explicit fields instead of asking it to infer everything from copy written for humans.

That is useful plumbing. It is not a paid placement, not a ranking signal anyone has shown you, and not a certificate of quality. Plumbing is worth having and worth doing correctly. It is not worth buying a promise attached to.

It is also worth being clear about the shape of the claim. These figures describe how often a measured configuration was present in one sample, on one set of dates, under one checklist. They are not a statement about how any individual site behaves in every circumstance, and a practice that scores badly on a scan like this may have a straightforward explanation that the scan had no way to record. That caution cuts against our own headline, which is why it belongs beside the headline rather than three sections below it. Treat the three-state split as a description of a sample, not a verdict on any site in it — including yours.

The experiment that shows the limit

Researcher Mark Williams-Cook placed a false address inside deliberately invalid structured data, while keeping the visible page text saying something different. ChatGPT and Perplexity repeated the false address back to him.

Two things follow, and only two. The assistants read a fact from the page and reproduced it, so they were clearly reading. And the markup being invalid did not stop them, so valid structured data was not acting as the gate that decided whether the fact was used.

What does not follow: that schema is useless, or that anyone should imitate the experiment. That test is not part of our dataset, it is a single published demonstration rather than a large study, and its lesson is bounded. It shows why "add markup and get cited" is too strong a claim. It does not show that markup is pointless, and it certainly does not license writing something in structured data that contradicts your page.

One platform figure, and what it cannot tell you

Just over half the sample was built on the same platform: WordPress accounted for 289 of the 527 sites, 54.8%. That fact is often used as the setup for a punchline about plugin defaults.

We are not going to deliver that punchline, because we did not cross-tabulate platform against schema state in this study. We can tell you the sample's platform mix and the sample's schema mix. We cannot tell you from this data that any platform caused any schema outcome, and quietly implying it would be exactly the move this post is arguing against.

How to audit your own site

Ask your developer for the structured-data output from the live homepage — the actual rendered output, not a screenshot of a plugin's settings screen. The settings screen shows what was requested. The output shows what a machine receives, and those diverge more often than anyone expects.

Then compare every field against the visible page. Name, address, phone number, business type, opening hours, and the relationships between pages should agree with what a patient sees. A machine-readable statement that contradicts the visible page creates a worse problem than having no markup: clean syntax carrying the wrong fact, stated with more confidence than the prose.

Record the current state before you change anything. If you intend to measure visibility afterwards, keep the same questions, the same engines, and the same run rules — otherwise a changed answer later cannot be attributed to the change, even cautiously. And if a vendor wants to charge for markup work, ask them to state in writing what it is expected to accomplish. "Your business identity will be machine-readable" is an honest deliverable. "You will be cited" is not one anybody can support.

What remains unknown

We have not measured whether adding or correcting schema changes AI visibility outcomes. We have not measured which schema type, if any, assistants prefer for dental practices. We have not measured whether the incomplete configurations in that 42.7% band behave differently from the empty ones. This audit supports an implementation finding and nothing beyond it.

Common questions

Will schema get my practice cited?
Nobody can promise that, and this study did not test it.

Is missing schema proof that a website is unreadable?
No. Visible HTML can still be read, as the false-address experiment illustrates. Schema makes identity and facts explicit rather than inferred.

Is having some schema good enough?
Our audit scored 225 of 527 sites, 42.7%, as present but incomplete. Whether incomplete is good enough is a question this data cannot answer — but a yes-or-no audit will not even show you which state you are in.

Should markup ever disagree with the visible page?
No. A structured field is not a safe hiding place for a claim a reader cannot see or verify.

Where can the audit method be checked?
The aggregated schema counts, the denominator rules, and the limitations are published at citevio.com/data.

About the author and disclosure

Muhammed Veysel Erin is the founder of Citevio. Citevio is a vendor, not a dental practice. Citevio ran the audit described above. The result is descriptive: it reports how often structured data was present in one sample on one set of dates, and it makes no claim that schema causes visibility.
Category: Cosmetic Dentistry, Marketing, Practice Management, HR and Accounting
You must be logged in to view comments.
More In Cosmetic Dentistry
2023 Dental Townie Choice Awards
2023 Dental Townie Choice Awards
December 2023
Aesthetic Criteria
Aesthetic Criteria
Dr. Jan Hajtó
CE Credits: 2
Facial Harmony and the Lips... Recorded Live at the Cad Ray Symposium.
Facial Harmony and the Lips... Recorded Live at the Cad Ray...
Dr. Laurence Rifkin
CE Credits: 2
Townie Casefile: Mint for More
Townie Casefile: Mint for More
December 2025
More In Marketing
Google Analytics 4: What Dentists Need To Know Now
Google Analytics 4: What Dentists Need To Know Now
Dr. David A. Wank
CE Credits: 1.5
Howard Speaks: Staying On Track in 2024
Howard Speaks: Staying On Track in 2024
by Dr. Howard Farran, DDS, MBA
February 2024
Poll: Marketing
Poll: Marketing
July 2026
What to Post: 25 Essential Social Media Posts
What to Post: 25 Essential Social Media Posts
Dr. John Syrbu
CE Credits: 1.5
More In Practice Management, HR and Accounting
Division of Duties in a Dental Practice, Part 4: The Finance Coordinator
Division of Duties in a Dental Practice, Part 4: The Finance...
Sandy Pardue
CE Credits: 1.5
Increasing Your Patient Acceptance Rates
Increasing Your Patient Acceptance Rates
Dr. Shalin Patel
CE Credits: 1.5
Poll: Human Resources
Poll: Human Resources
March 2024
The $2,500 Phone Call Most Practices Miss
The $2,500 Phone Call Most Practices Miss
by Flint Geier
May 2026
Total Blog Activity
997
Total Bloggers
13,451
Total Blog Posts
4,671
Total Podcasts
1,788
Total Videos
Dentaltown Magazine
August 2026
Dentaltown Magazine August 2026 Read Current Issue Read Archived Issues
Sponsors
Townie Perks