Measured, 6 September 2026Every page of both pilot sites failed.
Two Michigan municipal sites, 40 pages requested on each: a city of about 6,000 and a village of about 3,800. Both are on the 2028 date. Neither is a customer, so neither is named here.
78 of 78
pages scanned failed at least one WCAG 2.1 Level AA rule
35 of 79
PDFs opened had no structure tree, so a screen reader gets no headings
23
of those had no text layer at all, which tagging cannot fix
5 of 6
failing rules on one site were on every page, so they are one fix each
Why the numbers are good newsOn the smaller site, five of the six failing rules were on every page scanned. A rule that fails on every page is in the template, so it is one fix, not thirty-eight. On the other site the whole picture was a single theme-level contrast problem repeated 531 times. That is the shape of a municipal accessibility problem, and it is why a fixed fee works here when an hourly audit does not. The backlog that is genuinely large is the documents: 23 of the PDFs we opened had no text layer at all, and no amount of tagging repairs a scan of a piece of paper.
Both scans stopped at 40 pages, so these are partial site scans and every count above is over the pages actually scanned. A full crawl finds more, which is what the readiness report is for. The only exception hint that matched on either site was a year in a document URL, at low confidence. That is a hint that a document is old, not a finding that it is exempt.