— free accessibility checking for public entities

Español

About this tool

The DOJ set a standard and a deadline. What is missing is the capacity for small entities to meet it.

The problem

CISA's public registry of federal .gov domains lists 13,573 local-government entities, and the Census of Governments counts roughly ninety thousand local government units overall. The overwhelming majority serve populations under fifty thousand, which puts them on the 2028-04-26 deadline.

A professional accessibility audit of a single municipal site is commercially priced in the thousands to tens of thousands of dollars — routinely more than a small town's entire annual technology budget. The free tools that exist return undifferentiated violation lists that a part-time IT contractor cannot act on.

The baseline study

To find out how large the gap actually is, we drew a random sample of 80 local-government domains from the CISA registry using a published seed, and scanned each one's home page plus up to two linked pages.

Results, 25 September 2026 (n=80, seed 2026)
MeasureResult
Loaded successfully65 of 80
Would not load at all15 of 80 — 18.8%
With at least one blocking failure31 of 65 — 47.7%
With at least one finding of any severity51 of 65 — 78.5%
Clean on automated checks14 of 65 — 21.5%

These figures are a floor, not a conformance rate. Automated checking reaches roughly a third of the success criteria, and the scan was three pages deep. The true non-conformance rate is necessarily higher.

We report aggregate results only and do not name individual entities. Publishing a worst-offenders list would produce a news cycle and no remediation. Any entity can run the scan on itself and see its own results in full.

How the checking works

Rule checking uses axe-core, the open-source engine most commercial scanners are built on, driven through a real headless browser. A second pass reviews a screenshot of the rendered page for six failure classes a rule engine structurally cannot detect — meaningless alternative text, colour used as the only cue, images of text, reading order that does not match visual order, non-descriptive link text, and accessible names that omit the visible label.

Findings from that second pass are labelled as such and carry a confidence level. Verify them before acting.

Contact

Questions, corrections, and methodology critique are all welcome — particularly from people who use assistive technology daily. Email hello@muniaudit.com.

Check a site