How slow are dental websites? We measured 256 of them

Two hundred fifty-six dental practice websites. One simple test: how long before anything appears on the screen. 50.8% of them — more than half — took longer than three seconds. Does that explain why so many dental practices are invisible to ChatGPT and Perplexity? We measured the speed. We did not test the link between speed and visibility. Here's exactly what we found — and just as important, what we're not claiming.

Here's the framing question, up front, before any numbers: is website speed a factor in whether AI recommends a dental practice, or is it a completely separate issue? The honest answer is we don't know, and this video will not pretend otherwise. What we can tell you, with real data behind it, is how slow the typical dental website actually is — and why we could only get a usable reading on about half the practices we tried. That gap alone tells you something about how hard real-world data collection actually is.

First Contentful Paint — FCP for short — is the moment a browser first shows any part of a page: a headline, an image, even just a background color. It isn't a lab abstraction. It's the point where a patient looking at a blank tab finally sees something — before they've seen your practice's name, your hours, or your phone number, regardless of how polished the site turns out to be once it finishes loading.

We measured FCP with a single mobile reading from Google's own PageSpeed Insights tool — one reading per site, taken once, in June 2026. Worth flagging honestly: Google's own guidance recommends judging this across many repeat visits, not one snapshot. What we have is a real number, taken once, on a specific day — not a long-term average, and not a trend line.

We tried this on the same 527 dental practice websites from our broader June 2026 research. Only 256 of them returned a usable reading. The other 271 didn't fail a speed test — the tool simply never returned a usable number. Sometimes the request timed out, sometimes it hit a rate limit, sometimes the page just wouldn't render. We recorded those as pending, not as slow and not as fast, because that's what actually happened: nothing was measured. We genuinely don't know which way that unmeasured group leans, and we're not going to guess.

Here's what we found in the 256 we could measure. The median First Contentful Paint is 3.01 seconds — just past the line where Google's own guidance starts calling a page slow. And 50.8% of those 256 practices — more than half — took longer than three seconds before anything appeared on screen at all.

That's not a result buried out in the slow tail. That's the midpoint — the typical site. Line up all 256 practices from fastest to slowest, and the one standing right in the middle still takes just over three seconds to show anything. Half the group is slower than that.

Three seconds might sound trivial if you're used to fast internet on your own phone. It isn't trivial to a patient searching in pain, on a spotty connection, comparing three practices in three open tabs. The tab that shows something first often keeps the attention the tab that loads best never gets a chance to earn.

Now, the part we want to say clearly, and say twice. We are not telling you that a slow website causes a dental practice to disappear from ChatGPT or Perplexity. We haven't run a test that isolates speed as the cause of anything. This is a stand-alone measurement of how fast a page loads — nothing more, nothing implied. A correlation between two numbers in the same dataset wouldn't prove causation even if we had one, and right now we don't, because we haven't tested for it.

To actually test whether speed affects AI visibility, you'd need a controlled comparison: matched practices, same market, same content, different load times, then a fair read on which ones the engines mention. We haven't run that study. We're not aware of anyone who has published one with real numbers behind it. Until that exists, a claim that fixing your load time gets you recommended by ChatGPT is a guess dressed up as a finding — and we're not going to make that guess just to give you a tidier story.

We're spending more time on what we didn't measure than most research videos would. That's deliberate. A single confident number is easy to make sound bigger than it is. A number surrounded by its own limits is harder to spin into a sales pitch — and that's exactly the kind we'd rather give you. It's also less satisfying to record for a video, but satisfying and accurate aren't always the same thing.

So why measure this at all, if we're not claiming it matters for AI? Because a slow first paint is a real cost with or without AI in the picture. A patient staring at a blank screen for three seconds hasn't seen your hours, your phone number, or your name yet — and some of them will leave before any of it loads. That's a human-visitor problem on its own terms, worth fixing whether or not it ever touches an AI answer.

There are well-known reasons a small-business website tends to load slowly — heavy, unoptimized images, scripts that block rendering before any content shows, a page-builder template carrying a large plugin stack. Those are widely cited causes in general, but we did not measure image weight, script behavior, or plugin counts in this study, so we're not presenting them as our own finding here. What platform these sites run on is a separate question we've looked at elsewhere — not in this video.

We could have waited until we had a usable reading on all 527 sites before publishing anything. We didn't, because a majority reading is still real information as long as we're upfront about what's missing. Waiting for a perfectly complete dataset before saying anything true is its own kind of dishonesty — it just looks more patient.

For context: in that same broader research line, we separately found that 96.2% of the 527 practices — 507 of them — got zero mentions from ChatGPT or Perplexity. That's a different study, measuring a different thing: structured data, not page speed. We are not merging the two studies into one story. We have not tested whether the slow group and the unmentioned group even overlap, and saying they're connected right now would be exactly the kind of guess this video is refusing to make. The full study is linked below if you want it.

One more boundary, on purpose. This video does not connect page speed to reviews, to schema markup, or to anything else we've measured separately. Every number here stays inside its own study. Mixing findings across studies to build a bigger story is exactly the kind of shortcut that makes research untrustworthy, and we'd rather stay narrow and be right. We'd rather publish three narrow, defensible findings than one broad claim that doesn't hold up under a second look.

To recap: 256 of 527 dental practice websites gave us a usable speed reading. The median is 3.01 seconds. 50.8% of that group took longer than three seconds. We don't know about the other 271, and we're not claiming speed explains AI visibility — we're claiming it's slow, measured, and worth fixing on its own terms.

Every number in this video, the full methodology, and the raw CSV are open at citevio.com/data. If you think a figure here is off, tell us — we'd rather be corrected than quoted wrong. This is Citevio's third video: measured, dated, and open to checking, not a promise about rankings we can't deliver. Thanks for watching.