Scanner Report Triage, or Which Flags Are Real

What We Keep Finding

A scanner is an honest worker with a known flaw. It flags what looks wrong rather than what is wrong. The icon font reads as an empty link. The decorative image gets flagged for alt text it should never carry. The contrast warning fires on text sitting behind an overlay that no user ever sees. Your developers chase a few ghosts, quietly stop trusting the list, and the real failures buried in it stop getting fixed along with the fake ones.

There is a mechanism behind the argument you are about to have with somebody, and it is worth knowing before you have it. Take a form field labeled by nothing but its placeholder. The rules that work out a control's name put the placeholder in that list, so the field has a name, so it passes the scanner's label check and passes W3C's own test rule for the same thing. What it fails is a different criterion, the one about labels and instructions, and that criterion has no automated test rule at all. So your export says the form is fine, our auditor says it is not, and neither of us is being difficult.

Then triage says plainly what your export was never going to cover. We graded 356 of the 432 techniques and failures W3C publishes, and 10 can be settled by a machine outright. W3C's own automated test rules run out sooner than that. Of the 86 criteria in the standard, 49 have no test rule at all, and across the whole published set only four rules treat a pass as meaning the criterion is satisfied. On all the others, the wording W3C puts on its own rule pages says a pass means the criterion still needs testing. A clean scan was never the whole answer, and after triage you at least know which half of it you are holding.