Notes from the pipeline

What we learn running scans for other people

Short pieces on evidence, regressions and the boundary between measuring accessibility and claiming it. No newsletter, no gated PDFs.

FeaturedTesting7 min read

Why automated testing finds only part of the barriers

Automation is excellent at contrast ratios and missing names. It cannot judge whether alternative text is meaningful. Both matter.

Read
Key points
  • Automation reliably catches machine-checkable failures.
  • Focus order, screen-reader flow and meaningful alternatives need a person.
  • A credible report says which is which.

See what a scan says about a site you look after

Ten pages, mapped to EN 301 549 clauses, with the two most serious findings shown in full. The report lands the same day and nothing is installed on the site.