What 142 dental website audits actually found
Four practices scored above 70 out of 100. The sample skews toward the most visible practices in the market, which makes these findings the optimistic version.
Austin + Georgetown, TX · 142 practices · September to October 2026
Between September and October 2026, UberBrains audited 142 dental practice websites across Austin and Georgetown, Texas. Every site was scored out of 100 against the same checks: performance, analytics and measurement, security exposure, structured data, accessibility, and site hygiene.
Four of the 142 scored above 70. This is the first release from a seven city programme. Austin and Georgetown are complete.
Thirty-eight percent carry the structured data that lets a machine read what they are. The remainder are unreadable to the AI systems that increasingly answer patient questions before a search result is clicked.
We audited
the winners
The sample was drawn from the practices that surface. Which means it skews toward the ones already findable: better ranked, more actively marketed, more likely to have commissioned and maintained a website.
This is a sample of the practices that are winning. The median dental website in Central Texas is almost certainly worse than a 45.
The distribution
has no top end
Scores cluster between 30 and 69. That band holds 120 of the 142 practices. Above it, almost nothing. Four practices reached 70. None reached 75.
Practices per score band · n = 142
In a market of 2,350 practices, sampling the most visible ones, the strongest website found scored 72. There is no incumbent to displace. The category is open.
The cost is not abstract. Traffic arriving at these sites was paid for, through advertising, through search investment, through the directory listings the practice maintains. A page that takes 6.8 seconds to finish loading on a phone is spending that acquisition budget on people who leave before they see it.
One homepage in the sample loaded 532 images. Another took 24.9 seconds. These are not edge cases of ambition. They are sites nobody opened on a phone after launch.
n = 103 sites where load time was captured
A third of the market
cannot see itself
installed
collecting nothing
working
Twenty-two percent of sites measured had no analytics. Ten percent were running Universal Analytics, switched off by Google in July 2023 and collecting nothing since. Roughly one in three practices has no mechanism to answer the question where did my last ten patients come from.
They can still spend on marketing. They cannot evaluate it. Every advertising dollar, every referral, every form submission arrives unattributed, and next year’s budget is set on instinct.
This is the finding with the clearest commercial cost and the lowest cost to correct.
n = 107 sites where the analytics stack was captured
The front door
is open
49 of 59 sites checked
On 55 percent, the site published its own list of registered usernames through the default REST endpoint. A valid username and a reachable login form is half of a credential attack already completed. The remaining half is automated and cheap.
Dental practices hold patient records. The regulatory exposure is not theoretical. Sixteen percent of sites had a content security policy. Eighty-four percent had none.
n = 59 sites where exposure checks completed
Invisible to the machines
now answering the questions
38 of every 100
Two terms matter. AEO, answer engine optimisation, is being selected as the direct answer to a question. GEO, generative engine optimisation, is being woven into what an AI system produces when someone asks it to recommend a dentist. Both require a machine to be able to read the site. Sixty-two percent cannot be read.
n = 37 sites where schema was validated
The small things
nobody went back for
Six of 37 sites had an outbound link that went somewhere wrong.
- A Facebook icon pointing at a dead page
- A Yelp link opening a different practice, on Yelp’s Philippines site
- An Instagram icon going to instagram.com rather than the practice account
- Two different Facebook pages linked from one site
- An address link opening Google Maps in Springfield, Tennessee
None of these required skill to find. They required somebody to click.
n = 37
It is not a technology problem.
It is an attention problem.
The failures in this dataset are not sophisticated. A reachable login page. A dead analytics tag. A broken link. A slow homepage. Each is fixable in an afternoon by someone who looks.
A dental website is bought as a capital expense. It behaves like an operational one. The gap between those two assumptions is where every finding in this report sits.
Austin and Georgetown are complete at 142 practices. Five cities remain. When the Central Texas census is finished, we will publish the per city comparison and the full category breakdown.
We run two audits.
They answer different questions.
The instrument behind this report. It scores one website on what is technically true about it: how fast it loads, whether anything is measuring it, what it is exposing, whether a machine can read it, and what is quietly broken. You get a score, the findings ranked by what each one costs you, and the specific fix for each. It takes a day. Most of what it surfaces can be resolved in a week.
The larger question. Not how your site performs, but whether it is winning. It maps the practices you are actually losing patients to, what they are visible for that you are not, where they are being chosen by search and by AI, and which of those gaps is worth closing first.
The first tells you what is wrong. The second tells you what it is costing you, and against whom.
Recent Comments