Why we print our own coverage number
The first thing a customer sees on most accessibility scanners is a score. Ninety four out of a hundred. A green ring. A letter grade. It is a good number, and it is measuring the wrong thing.
What it measures is the rules the tool happened to run. Not the standard. Automated testing reaches a definite verdict on a minority of the 55 WCAG 2.2 Level AA success criteria. A tool that runs forty rules cleanly and reports 98% has told you about forty rules, and implied something about fifty-five criteria.
What we do instead
Every report carries a coverage figure: how many criteria this scan actually decided, out of the full set. A criterion we could only partly test is never counted as a pass. Below it there is a list, by name, of every criterion a person still has to check, and one sentence on what checking it involves.
It is a worse number. A first-time reader sees something like 38% and their reaction is not delight. It is the true number, and it is the one that tells them what work is left.
The commercial argument
We are not printing it out of virtue. A buyer who has been sold a 98% and then receives a demand letter does not come back. A buyer who was told, on day one, exactly where the line was is a buyer who trusts the next thing we say. The second one is the business.