Tell a company its website is slow and the usual answer is that everyone's is. It is a fair defence and until now nothing but an impression could challenge it. So we measured it.
In August 2026 we took 386 websites of companies, municipalities, schools and organisations in the Zlín region, ran them through one tool in a single week and read what each page loads by itself. No names, no ranking, only aggregates. This article retells the findings; the full report with methodology and limits lives on the site and its numbers are computed live, so they cannot be copied wrong.
Nine sites in ten sit below the line
344 sites landed below ninety points, that is 92 %. 281 of them, 76 %, are below seventy, where speed measurably puts people off. Twenty eight sites in the whole region scored above ninety.
This is not one number on a chart. The median time until the main content shows on a phone came out at 9.2 seconds. That is a lab measurement on a throttled connection, a simulation of a phone on the move rather than office wifi. But that is exactly how customers arrive: in a car, on a site, on the way back from a meeting.
How mobile performance came out
| Label | Value |
|---|---|
| 90 points and above | 8 % of sites |
| 50 to 89 points | 76 % of sites |
| below 50 points | 16 % of sites |
A high score does not mean a good website
This is where a role model was supposed to appear. It did not. We looked at what sets the fastest eight percent apart from the rest and the answer is: there is almost nothing on them.
Sites scoring 90 and above: a median of 193 kB and nine downloaded files. Four in five run on no recognised system at all. The group holds static pages built years ago and pages with a single heading and a phone number.
The rest of the sample: a median of 3,014 kB and 64 files. Sixteen times heavier and seven times more files, but also content, photos, forms and shops.
Speed is a condition, not a goal. An empty page with ninety points answers not a single question a customer has. What matters is a site that is fast and has something to say, works on a phone and meets the legal minimum. There are few of those in the region.
What it runs on does not decide the speed
The obvious suspect is WordPress. The data does not support it. Median performance differs between systems by a few points: WordPress 61, Joomla 63, Drupal 60, sites with no recognised system 62. The extremes belong to groups of a handful of sites where a single number means nothing.
Speed is not decided by the choice of system but by who built the site and whether anyone maintains it. The same WordPress can run briskly or crawl, depending on how many plugins were piled onto it and whether anyone updates them.
What is not done, even though it costs nothing
The strongest findings are not about speed. They are about things set once, at no extra cost, that are missing anyway:
- Forced secure connection. Almost everyone has https, but only 76 sites force it, a fifth. The rest trust the browser to work it out.
- Security headers. 263 sites, seven in ten, do not have even two of the basic ones. It is one line in the server config.
- Zooming on a phone. 104 sites still block it, so anyone with weaker eyesight simply cannot enlarge the text.
- Image descriptions. Missing on 125 sites. Without them a screen reader cannot read the content and search engines understand it less well.
- Consent for tracking. 80 sites run advertising trackers without asking anything at all. That is not a server setting, that is a fine waiting for a complaint.
Accessibility fared better than expected: a median of 85 points. Yet only 115 sites out of the 372 scored cleared ninety, under a third, and since June 2025 accessibility is an obligation for a large share of businesses, not a bonus.
The bar is low. That is not a reason to sneer, it is an opening: a site that loads fast, reads well on a phone and meets the legal minimum already puts you in the top few percent of the region
What to do about it as an owner
You do not have to believe us or your supplier. Measure the site yourself, it takes half a minute and Google does the measuring, not us. The tool sits on the page about choosing a supplier together with a guide to vetting a web supplier by their own website.
And if you want to know how your own site did, you can look it up in the report. We send the analysis to an email address in your own domain, because it is a finding about your company and belongs to nobody else.
Will you publish how a particular site did?
No. Only aggregates go out, no names and no ranking. A company can ask for its own result and we send it to an email address in the same domain as the measured site. A competitor, a journalist or a passer-by cannot get to someone else's result.
How did you choose which sites to measure?
The sample came from a public company directory for the Zlín region, topped up from public registers. It is not a representative sample of the region's economy: companies that make themselves known are over represented, so the more active ones. A random sample would probably have come out worse.
Is this not just an ad for your services?
We build websites, so we do have an interest, and it is fair to say so out loud. That is why the report carries its methodology and a list of what the measurement cannot do, and why every number is computed from the database rather than typed into the text. Anyone can re-measure with the same public Google tool and arrive at the same result.
Does a low score mean the site does not work?
No. A score is a measure, not a verdict. A slow site can still bring in work, it just brings in less than it could and some people leave before the page loads. What matters more than one number is how it combines with the others: speed, accessibility, the legal minimum and upkeep together.