Why automated accessibility checkers aren't enough
Automated tools are useful and fast — but they catch only a fraction of real accessibility barriers. Here is the gap.
Automated accessibility checkers are genuinely useful. They’re fast, cheap, and great at catching the obvious, repetitive problems. But if a tool tells you your site is “98% accessible,” it’s measuring the wrong thing — because most of accessibility can’t be checked by a machine.
What tools are good at
Automated scanners reliably catch things like missing alt attributes, empty form
labels, and low color contrast. Run one regularly — it’s a great first line of defense
and a sensible CI check.
What they miss
The catch: studies consistently find automated tools detect only around a third of WCAG issues. The rest need judgment:
- Is the
alttext actually meaningful, or just present? - Does the focus order follow the visual flow?
- Does that custom dropdown work with a keyboard and screen reader?
- Is an error message announced, or only shown in red?
A tool can confirm a label exists. It can’t tell you the label makes sense.
Use both, in the right order
The strongest approach is layered: automated checks to catch regressions continuously, and human testing to find the barriers that actually stop people. One without the other leaves a false sense of security — usually the expensive kind.