Slower than a patient would probably wait for. We scanned 527 US dental practice websites in June 2026 and completed a First Contentful Paint reading on 256 of them. The median is 3.01 seconds, just past Google's own threshold for a "poor" score. We could not read the other 271, and we do not know whether they are faster or slower.
First Contentful Paint, or FCP, is the time from when a browser starts loading a page to when any part of its content (text, an image, a background color) actually appears on screen. It is a real-world number, not a lab abstraction: a patient staring at a blank tab has not yet seen your practice's name, hours, or phone number, no matter how good the page looks once it finally renders. We measured page speed on its own, as one technical fact about a website. What speed does and does not explain about whether a practice shows up in ChatGPT or Perplexity is a separate question, and we answer it directly further down this page.
How slow is the average dental website?
On the median site we could measure, content starts appearing at 3.01 seconds. A quarter of the sites we read beat 2.12 seconds; a quarter took longer than 4.27 seconds, and the slowest tenth took 6.21 seconds or more.
| Percentile | First Contentful Paint |
|---|---|
| Fastest (min) | 0.79s |
| 25th percentile | 2.12s |
| Median | 3.01s |
| 75th percentile | 4.25s |
| 90th percentile | 6.61s |
| Slowest (max) | 13.62s |
| Threshold | Sites past it | Share of 256 |
|---|---|---|
| Past 2.5 seconds | 161 | 62.9% |
| Past 3.0 seconds (Google's "Poor" line) | 130 | 50.8% |
| Past 4.0 seconds | 75 | 29.3% |
Source: Citevio dental practice web anatomy study, June 2026, n = 256 of 527 practices scanned in 7 US metros plus 9 smaller nearby towns. Full dataset and CSV: citevio.com/data.
What does "First Contentful Paint" actually measure?
It is one of the metrics Google uses to describe how a page feels to load, and Google publishes three bands for it: 1.8 seconds or less is Good, 1.8 to 3.0 seconds is Needs Improvement, and anything past 3.0 seconds is Poor. Sorted into those bands, 43 of the 256 practices we measured, 16.8%, land in Good. Another 83, 32.4%, land in Needs Improvement. The remaining half, 130 practices, 50.8%, land in Poor, including the median site at 3.01 seconds.
| Google's band | FCP range | Practices | Share of 256 |
|---|---|---|---|
| Good | 0–1.8s | 43 | 16.8% |
| Needs Improvement | 1.8–3.0s | 83 | 32.4% |
| Poor | Over 3.0s | 130 | 50.8% |
Google's guidance recommends judging FCP at the 75th percentile across repeat visits from real users, blending mobile and desktop. Our numbers are not that: each is a single mobile Lighthouse reading through Google's PageSpeed Insights API, taken once, in June 2026. Treat the comparison to Google's bands as a useful yardstick, not an identical measurement.
One honest caveat: FCP is not one of Google's three official Core Web Vitals. Those are Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift, the metrics Google says matter most for user experience and search. Google's own documentation describes FCP as a supplemental, diagnostic metric, useful for spotting render-blocking problems on the way to a slow Largest Contentful Paint, rather than a ranking signal in its own right. We report it here because it is what our scanner measures, and because "when does anything appear at all" is still a real moment for a patient waiting on a page, just not the specific number Google weighs most heavily.
Sources: web.dev, "First Contentful Paint (FCP)" and web.dev, "Web Vitals", Google, both checked 26 July 2026.
Why could you only measure 256 of 527 practices?
Each reading needs a live call to Google's PageSpeed Insights API for that specific site, and on 271 of 527 sites that call did not return a usable number: a timeout, a rate limit, or a page PageSpeed could not render. We recorded those as pending, not as a failed or slow reading, because that is what actually happened. Nothing was measured.
This matters for how you read the 50.8% figure above. The 271 unmeasured practices are not the slowest sites in our sample with their scores hidden. They are sites where the tool never got a number back, for reasons that have nothing to do with how fast the page loads. We do not know whether the missing group is faster or slower than the group we could measure, and the direction of that gap is not something we are willing to guess. If you want the fully reconciled dataset, including which fields are complete and which are pending, it is free to download at citevio.com/data.
Does a slow website cause a dental practice to disappear from AI answers?
We don't know, and this page does not claim to. We measured page speed as a stand-alone technical fact. We did not test whether it changes whether ChatGPT, Perplexity or Gemini recommend a practice. That would take a controlled comparison we have not run, and a correlation between two numbers in the same dataset would not prove one either.
What we have measured, separately, is a mechanism that does gate AI access directly: a robots.txt file that blocks a crawler outright stops that crawler from reading the page at all, which is documented behavior rather than an inferred pattern. We cover that study on its own page. A slow first paint is a real cost. It is a bad first few seconds for a patient who did click through, and a signal search engines have used for ranking pages for years. But on the data we have today, we cannot tell you it keeps a practice out of an AI answer, and we are not going to imply otherwise to make this page's finding sound bigger than it is.
Related: are dental websites blocking AI crawlers? (a mechanism we did measure directly).
How fast should a dental website load?
Under Google's own bands, aim for 1.8 seconds or less to land in "Good." Only 16.8% of the practices we measured get there. Half take longer than 3 seconds, which is where Google's guidance starts calling a page "Poor," so a practice beating 1.8 seconds would already be ahead of roughly four in five of the sites we measured.
- Check your own number first. The checker below reads a live First Contentful Paint value for your site in seconds, alongside the other checks on this page.
- Large, unoptimized images and render-blocking scripts loaded before any content are widely cited as common causes of a slow first paint on small-business websites generally — we did not measure image weight or script-blocking directly in this study, so we are not claiming that as our own finding here.
- A page builder or template with a heavy plugin stack can add meaningful load time before your practice's own content shows up. That's a platform question we cover in what platform do dental websites run on.
If you'd rather see the whole picture at once, the checker below reads your site and scores its AI readiness in seconds, speed included.
Prefer to question the engines by hand? See how to check what ChatGPT says about your practice.
How we measured this
The sample is 527 US dental practice websites in Charlotte, Austin, Raleigh, Nashville, Columbus, Tampa and Salt Lake City, plus 21 practices in 9 smaller nearby towns, scanned between 6 and 26 June 2026. For each site we requested a mobile-strategy report from Google's PageSpeed Insights v5 API and read the "First Contentful Paint" value out of the Lighthouse audit it returns. That single-pass request either succeeded, in which case we kept the number, or it did not, in which case the field is marked pending. We did not retry failed calls or average repeat readings, so this is one snapshot per site rather than Google's own recommended 75th-percentile-of-repeat-visits measurement.
- A snapshot, not a trend. Every number here carries the June 2026 scan window; a rerun on a different day will produce different figures, especially for sites mid-redesign or mid-migration.
- One reading per site. Real-world load time varies visit to visit. We report a single measurement, not a percentile across repeat loads the way Google's own guidance recommends.
- Partial base. The reading completed for 256 of 527 practices. The direction of the gap in the other 271 is unknown; we do not claim the missing sites are faster, slower, or typical.
- Sample, not census. This is 7 metros plus 9 nearby towns, not a national sample. We publish a national figure only once we can read at least 2,000 sites; this study does not meet that bar.
The full framework behind our scans, scores and city reports is in how we measure AI visibility, and the underlying dataset is free to download at citevio.com/data. If you think a number on this page is wrong, email contact@citevio.com with the domain and what you saw. We would rather be corrected than quoted wrongly.
Common questions
How slow are dental websites?
In our June 2026 study we completed a First Contentful Paint reading on 256 of 527 US dental practice websites. The median is 3.01 seconds, which is just past Google's own threshold for a "poor" score (anything over 3.0 seconds). 50.8% of the practices we could measure took longer than 3 seconds, and 29.3% took longer than 4.
What is a good First Contentful Paint time?
Google's web.dev guidance sets three bands: 1.8 seconds or less is Good, 1.8 to 3.0 seconds is Needs Improvement, and anything over 3.0 seconds is Poor. In our sample, the practice at the 25th percentile (2.12 seconds) sits in the Needs Improvement band, and the median practice (3.01 seconds) is just inside Poor.
Why wasn't every practice's speed measured?
Each reading requires a live call to Google's PageSpeed Insights API, and on 271 of 527 sites that call did not return a usable result, so the field was left pending rather than marked as a failure. We do not know whether those sites are faster or slower than the ones we could read, and we are not willing to guess which.
Does a slow website cause a dental practice to disappear from AI answers?
We don't know, and this page does not claim to. We measured page speed as a stand-alone technical fact and did not test whether it affects whether ChatGPT, Perplexity or Gemini recommend a practice. That would need a controlled comparison we have not run. Separately, we have measured mechanisms that do gate AI access directly, such as a robots.txt file blocking a crawler outright.