Research question
As the extended ADA Title II deadlines approach, what share of state and local government documents would satisfy WCAG 2.1 AA, and is that share actually improving quarter over quarter?
Overview
The Department of Justice extended the Title II compliance dates to April 2027 for larger public entities and April 2028 for smaller ones, while stating that document remediation obligations remain in force under existing rules. Nobody publishes a recurring measurement of whether public documents are actually getting more accessible as those dates approach.
This index samples public documents from state and local government websites each quarter, runs the same published battery of automated accessibility checks on every document, and reports readiness by state, sector, and document class, with the method and its limits stated on every page.
Automated checks cannot establish conformance, and the index will say so prominently. What they can establish, at scale and consistently, is the floor: documents that fail machine-checkable requirements cannot be conformant, and the trend of that floor is the story.
Why this matters
Deadlines without measurement drift
A one-year extension becomes a two-year extension when nobody is watching the trendline. A public, quarterly index creates the accountability signal that budget owners respond to.
Agencies need a peer baseline
A county web team asking for remediation budget is far more persuasive with sector data: where their document readiness stands against peers, and how fast the gap is closing elsewhere.
What we measure
- Machine-checkable failure floor
- Share of sampled documents failing automated checks that any conformant document must pass
- Tagging presence and quality
- Untagged share, tag coverage ratios, and structure-tree quality signals
- Trend
- Quarter-over-quarter movement per state and sector with stable methodology
- Exposure
- Document volume estimates so percentages translate into real backlog sizes
Method
Sampling frame: official state and local government domains, stratified by state, entity size band, and sector, with the frame published.
Acquisition under the conversion pilot's provenance rules: official URLs only, recorded hashes, malware scanning, no redistribution.
Scoring: the published automated audit battery, versioned; every quarter runs the same version or discloses the change.
Publication: quarterly, with per-stratum sample sizes and confidence intervals, and a standing correction policy.
Evidence so far
Only recorded findings appear here. Anything not listed has not been measured yet.
The audit battery exists and is public
The structure, alternative text, table, and text-extraction checks that power the index already run in production and are documented with the benchmark corpus.
Demand signal
The most-cited industry study in this space is a single 2025 scan of 644,854 PDFs. It is cited constantly because nothing recurring exists. A quarterly index fills that vacuum.
Questions we expect
Can automated checks say a document is compliant?
No, and the index never will. Automated checks establish a failure floor: documents that fail them cannot be conformant. That floor, measured the same way every quarter, is what the index tracks.