Manual accessibility checks: when and why
The 30% of accessibility issues automation can't find for you.
Updated
Automated tools catch about 70% of WCAG issues. The other 30% need human judgment β and that's where manual checks come in.
Examples of what's manual
- Alt text quality β automation can flag a missing
alt, but only a human can tell whether "image123.jpg" is meaningful or not. - Reading order β does the DOM order make sense for screen readers?
- Label clarity β "Submit" vs. "Submit your application" β both pass automation, only one is helpful.
- Focus order β pressing Tab repeatedly should walk through the page in a logical sequence.
How manual checks work in HelloCivic
- Open ADA Scanner β Manual checks.
- Pick a site and a page. We surface candidate elements that need human review.
- For each candidate, mark it Pass / Fail / N/A and add a note.
- Results merge into the same per-page report as automated findings.
Best practice
Run automated scans every week and a manual review once per release. Manual checks compound β once you've labeled an element, future scans skip it unless the markup changes.
Was this article helpful?
Your feedback helps us improve our docs.

