Study: 13 Service Websites Under the Microscope and What Is Actually Broken
The received wisdom about small business websites is that “the SEO basics are not set up”: missing headings, empty descriptions, images without captions. We decided to check that with measurements rather than by ear.
We took thirteen real service business websites and captured the same set of metrics from each. The result was the opposite of expectations: the basic markup is in order almost everywhere, and something else entirely is broken.
How we measured
Being honest about method so the numbers can be judged soberly:
- Sample: 13 service business websites collected from search results for commercial queries.
- The homepage was captured, one request from one location, a standard desktop browser identifier.
- Measured: server response time, HTML weight, image count and share without captions, number of first-level headings, page title and description length, presence of the mobile meta tag, number of external scripts, presence of structured data.
- Two sites out of fifteen refused the request and are excluded from the counts.
This is a small sample from a single segment and makes no claim to statistical representativeness. But it is enough to test a widely repeated claim.
What turned out to be fine
| Metric | Result |
|---|---|
| Mobile meta tag | 13 of 13 |
| Page title present | 13 of 13 |
| Exactly one h1 | 12 of 13 |
| Structured data | 11 of 13 |
| Page description present | 11 of 13 |
| Images with alt text | 463 of 495, or 94% |
The idea of “small business sites with no SEO basics” did not hold up. A mobile version is accounted for everywhere, headings are in place, structured data is present on the large majority. The problems everyone talks about are mostly already closed.
What turned out to be broken
| Metric | Median | Minimum | Maximum |
|---|---|---|---|
| Server response time | 1380 ms | 560 ms | 2640 ms |
| HTML weight | 124 KB | 59 KB | 410 KB |
| External scripts | 15 | 2 | 45 |
The headline figure is the first one. The benchmark for server response is under 200 milliseconds. Not one site out of thirteen met it. Four came in under 700 milliseconds; three took longer than a second and a half.
It matters that this happens before the browser has downloaded anything at all. A second and a half went by before the first byte of markup. Scripts, fonts and images come after that, and every recommendation about optimising pictures operates on top of that delay.
Scripts: the second finding
A median of fifteen external scripts on a homepage. One site had forty-five.
Each script is a separate request, a separate domain to connect to, and a window in which the browser may not respond to a tap. That is usually where poor interface responsiveness scores come from, rather than from the heavy images people tend to blame.
Tellingly, the site with the most scripts in the sample was also the slowest to respond and the only one with two h1 headings instead of one. Problems generally travel in groups.
What follows from this
The gap between what gets discussed and what actually slows sites down looks like this:
- What gets discussed: image alt text, headings, meta descriptions. In this sample those are 90-100% covered.
- What is actually broken: server response time and third-party script count. Nobody managed those.
The reason for the gap is simple. The first group is fixed by a content manager in an hour and appears on every checklist. The second requires decisions about hosting and about which marketing tools stay on the site, and those are not the implementer’s call.
One more observation: nine of thirteen sites use lazy loading for images. So image optimisation was considered. But with a server response of a second and a half, the gain from it disappears.
Limitations of this study
So the numbers get used honestly:
- Thirteen sites is a small sample, an indicator rather than statistics.
- One market segment; other niches may look different.
- One measurement from one location: response time depends on load at that moment and on geography.
- Only the homepage was measured. Inner pages are often slower.
- HTML weight is not total page weight: images, fonts and scripts load on top.
How to repeat this on your own site
Three checks that give you the same figures for your site:
- Server response time. Browser developer tools, network tab, the first document request. Look at the wait before the first byte.
- Number of external scripts. Same tab, filtered by type. Count them and ask, for each, what it does and whether it is still needed.
- HTML weight. The size of that first request in the same tab.
If your server response exceeds a second, that is the most expensive problem on the site, and it is where to start rather than with a basic markup checklist. In our sample that is precisely where every single site fell short, while everything small business usually gets criticised for had long since been done.
