Public research report
The Website Quality Report
A scan of 15,542 public websites found widespread issues in performance, mobile experience, accessibility, structure, and overall site quality.
Executive summary
What this report shows
Most business owners judge a website by how it looks.
This report looks at the parts they usually cannot see: performance, mobile experience, accessibility, structure, and overall site quality.
The scan does not prove why any individual website performed poorly. It does not prove who built the site, what was paid for it, or whether the site produces business.
It does show a broad pattern: many public websites are missing or underperforming on basic quality signals that affect how fast they load, how well they work on mobile, how accessible they are, and how structurally healthy they appear.
Key findings
What the scan found
Most business owners judge a website by how it looks. But the parts that affect speed, usability, accessibility, search visibility, and structure are usually hidden. This report looks under the surface.
Overall quality
Site Score distribution
Completed scans were concentrated between 50 and 70. Very few completed scans reached the highest scoring range.
Only 49 out of 15,541 completed scans reached a Site Score of 90 or higher.
Failed Scan
Critical
Poor
Needs Work
Fair
Good
| Score range | Websites | Share | Distribution |
|---|---|---|---|
| Failed Scan | 1 | - | |
| 0-30Critical | 128 | 0.82% | |
| 30.01-40Critical | 673 | 4.33% | |
| 40.01-50Poor | 1,949 | 12.54% | |
| 50.01-60Poor | 4,761 | 30.64% | |
| 60.01-70Needs Work | 5,305 | 34.14% | |
| 70.01-80Needs Work | 2,262 | 14.56% | |
| 80.01-90Fair | 426 | 2.74% | |
| 90.01-100Good | 37 | 0.24% |
Performance
Load time distribution
Load time was one of the clearest issues in the completed scan results. The average measured load time was 7 seconds.
62.60% of completed scans took more than 5 seconds to load.
25.38% took 10 seconds or longer.
Under 5 seconds
5-10 seconds
10+ seconds
| Load time | Websites | Share | Distribution |
|---|---|---|---|
| Under 2 seconds | 1,882 | 12.11% | |
| 2-3 seconds | 1,125 | 7.24% | |
| 3-5 seconds | 2,792 | 17.97% | |
| 5-10 seconds | 5,799 | 37.32% | |
| 10+ seconds | 3,943 | 25.38% |
Accessibility
Accessibility issue distribution
Accessibility issues were common in the point-in-time scan results. These measurements identify detectable issues, not the complete lived experience of every visitor.
77.74% of completed scans had at least one accessibility issue.
21.96% had 10 or more accessibility issues.
0 issues
1-5 issues
6-10 issues
11-20 issues
21+ issues
| Detected issues | Websites | Share | Distribution |
|---|---|---|---|
| 0 issues | 3,444 | 22.18% | |
| 1-5 issues | 6,737 | 43.40% | |
| 6-10 issues | 2,260 | 14.56% | |
| 11-20 issues | 1,778 | 11.45% | |
| 21+ issues | 1,306 | 8.41% |
Mobile and structure
Hidden quality signals were consistently weak
Mobile experience and structure health are separate from visual design. These measures help describe whether pages are fast, readable, and technically sound.
Mobile score under 70
9,814 completed scans had a mobile score below 70.
Structure health under 70
4,924 completed scans had structure health below 70.
Performance score under 70
Performance gaps can exist even when a website looks acceptable on the surface.
Critical
Poor
Needs Work
Fair
Good
| Structure health range | Websites | Share | Distribution |
|---|---|---|---|
| 0-30Critical | 0 | 0.00% | |
| 30.01-50Poor | 944 | 6.10% | |
| 50.01-70Needs Work | 4,346 | 28.07% | |
| 70.01-90Fair | 10,153 | 65.59% | |
| 90.01-100Good | 40 | 0.26% |
Methodology
How to read the results
This dataset summarizes a point-in-time scan of 15,542 public websites. Scores should not be interpreted as permanent fixed values. Live website measurements can change due to server response, network conditions, third-party scripts, redirects, caching, temporary outages, and site updates.
Downloads
Download the dataset and supporting files
Download links point to source workbook exports in assets/data/,
including a public dataset with business names and phone numbers removed.