Nobody sets out to spend money badly on accessibility. The waste comes from a handful of beliefs that sound sensible, get repeated by people who ought to know better, and go unchallenged until somebody pays to undo the consequences. Here are the six we meet most often, and what each one costs.
1. A Clean Automated Scan Means the Site Passes
A scanner reports what it can measure, and that is a smaller slice of the standard than the report implies. W3C publishes no conformance-tested automated rule at all for 24 of the 55 Level A and AA success criteria, and by our own technique-by-technique count a machine fully settles 10 of the 356 WCAG techniques and documented failures we graded, which is 2.8%. Tools are excellent at contrast ratios and missing alt attributes. They cannot judge whether meaning arrived.
The clearest illustration is a form field labeled by a placeholder instead of a label. That field still gets an accessible name from the placeholder, so your scanner's label check passes it. What it fails is the rule about labels and instructions, which has no automated test rule behind it at all, and the label vanishes the moment somebody starts typing. This is the single commonest reason a clean scan report and a human tester disagree with each other.
The bill for this myth arrives as a false sense of finished. Owners stop work, publish an accessibility statement, and find out the truth when a customer cannot complete a purchase. What automation can and cannot catch is set out rule by rule in automated versus manual testing.
2. An Overlay Makes the Site Compliant
Overlays do repair some things, and the field's own reference document on them says so, which is why we are not going to tell you they fix nothing. What they cannot settle reliably is the part that decides whether a site works. Text alternatives, form labels and error handling, and keyboard access. A script can rename a button in the page it is running on, and we are not going to pretend otherwise. What it cannot do is work out what the button should be called, and it never touches the markup your server sends, so the defect a complaint would cite is still in your codebase either way.
The protection claim is the one worth testing, because protection is why most owners buy. UsableNet's tracker counted 4,928 web accessibility lawsuits filed in 2025 across federal courts plus the New York and California state courts, which is the scope it says it reads. The same tracker publishes a number nobody selling a widget quotes. Every month it counts the defendants who were sued while running a third-party accessibility control, and in July 2026 that was 113 of 401. A widget on the page did not stop those cases being filed. The cost of this myth is a subscription fee sitting on top of the original unsolved problem. Our overlays page has the documented record.
3. Accessibility Only Matters if Somebody Complains
This treats accessibility as a legal risk rather than as access to a group of customers, and it gets the economics backwards. People who cannot use your checkout mostly do not complain. They leave, and you record their departure as a bounce.
That is exactly why this myth survives. The cost is invisible, because lost sales show up in your analytics as conversion you never had rather than as a bill somebody hands you. A complaint is the expensive version of information you could have bought cheaply.
4. It is a Developer Problem
Look at what actually gets found. In WebAIM's February 2026 survey of a million home pages, the two commonest detected failures were low contrast text, on 83.9 percent of pages, and missing alt text, on 53.1 percent. A color choice and a missing sentence. Neither is a developer's decision, and six error categories account for 96 percent of everything automated tools found.
Link text that describes nothing, headings chosen for their size, images labeled with filenames and PDFs exported without tags all arrive through the content workflow, after the build is finished. The bill arrives when you pay a developer to fix something a content editor reintroduces the following week. Who owns which half is mapped out in the content editor's checklist.
5. We will Do It After the Redesign
This one sounds like planning, and it almost always ends with the redesign shipping the same faults plus a few new ones, because nobody wrote accessibility requirements into the brief. The moment to specify it is before anyone writes code, in concrete testable language a supplier can price.
The cost of waiting is a second remediation project on the site that was supposed to fix the first one. What to put in the brief is in briefs and contracts.
6. Accessible Means Ugly
WCAG asks for contrast ratios, a visible focus indicator, a heading structure that matches the document, labeled form fields, and text that survives being enlarged. Read the standard end to end and you will not find a typeface, a palette, a layout or an opinion about taste anywhere in it. Some of the most disciplined design on the web clears Level AA without trying.
The cost of this myth is the smallest of the six and it is the most common. It usually shows up as a focus ring somebody removed because it clashed. That is one line of CSS to put back, and it is one of the failures we report most often.
The pattern underneath all six
Every one of these gives somebody permission to stop work early. That is what makes them attractive, and it is why the bill always arrives later than the decision to stop. If you want the cheap version of this information, the free 10-page scan covers the machine-checkable part today and tells you what it could not see.