Improved duplicate error detection
Index CSGOV uses AXE and HTML Code Sniffer to measure accessibility. A certain portion of the reported errors is shared between them. Until now we mistakenly included them in the score calculation twice. Duplicate errors are now counted only once, but you will still see both in the detailed error list.
Measurement tooling update (Lighthouse 13, Pa11y 9)
We have updated the measurement tools to their current versions: Pa11y 9 with a newer version of axe-core (accessibility measurement), Lighthouse 13 (performance measurement).
The necessary components of the measurement infrastructure have also been updated (Puppeteer 24 and Node.js 24). All tests now run in a single Chrome browser on the CI runner, which unifies the environment for accessibility and performance measurements as well as screenshots.
Newer tool versions may change the measured values: updated axe-core rules may reveal errors that earlier versions did not detect, and Lighthouse 13 adjusts the performance score calculation. A website's score may therefore change even without any modification of the measured website.
What Has Changed
Several errors have been reclassified from critical to moderate, and therefore no longer affect the score. These changes affect approximately 10% of websites.
Extended structural inspections and health checks of the template
We have added advanced structural checks to the measurement that standard automated tools (Axe, HTML CodeSniffer) do not run by default:
- Clickable target size (WCAG 2.5.8) — whether buttons and links are large enough for comfortable and safe interaction (min. 24 × 24 px).
- Reflow at 320 px width (WCAG 1.4.10) — whether the content requires horizontal scrolling on narrow displays (mobile, zoom).
Therefore, a new section titled “Template Structural Health” has been added to the measurement details, consisting of two parts:
- Advanced Checks — results of the checks listed above; these are tracked separately.
- Recurring Errors Across Pages — critical and severe issues that appear on multiple tested pages. Such errors are typically found in the template, so fixing them in one place will improve the entire website.
These checks are diagnostic. They do not change the score calculation, its values, or the ranking—the score and all historical measurements remain unchanged. The goal is to highlight structural barriers that the score itself does not capture and to facilitate their correction.
English translation for the whole website.