Original research · Vertical · July 2026

The State of Government CMS Readiness 2026

Public information is meant to be the easiest thing on the internet to find. That is more or less the entire point of it. So we measured every reachable US government homepage across 16,314 .gov and .mil domains (9,324 live and measurable) against the eleven signals that decide whether a site stays findable through a replatform. Government came back the worst-performing segment we have measured anywhere: the lowest grades, the least structured data, and the widest gap on the one signal that now decides whether an AI assistant will quote a page at all.

The AI-visibility gap widens the closer you get to government

Read across three populations we’ve measured this year and the pattern is unmistakable: the more public and authoritative the content should be, the less machine-readable it actually is. Government sits at the bottom of every column.

Government is nearly universally on HTTPS (99.7%) and gets page titles right (91.9%) — the basics are in place. What is missing is the structured, answer-engine-ready layer: schema markup, canonical tags, and clean metadata. Those are exactly the signals a CMS migration drops on the floor when nobody is measuring, and they are the difference between an agency being named in an AI answer and being skipped.

Grades: two in three government sites are D or F

Scored across all eleven signals, 68.5% of measured government homepages land at grade D or F — far worse than the open web (42.3%) or higher education (42.6%). The long tail of county, municipal and special-district sites, often maintained with little budget, pulls the whole segment down.

What government runs — and the Drupal risk underneath it

Where a platform is detectable, WordPress leads and Drupal runs at 2.1× its web-wide rate — government is, after higher ed, Drupal’s second home. That matters because of what is running underneath: of the 603 government Drupal sites that expose a version, 15.8% (95 sites) are on end-of-life Drupal 7, 8 or 9 — unsupported, unpatched public infrastructure. Drupal 7 alone is the single largest version bucket among the aging tail.

Version is only observable on the sites that publish a generator tag, so the true end-of-life count is a floor, not a ceiling.

Readiness by platform

Every CMS with at least fifteen measurable government sites, by overall readiness and the two signals most often lost in a migration, structured data (SEO / AI visibility) and security headers.

Method and limitations

• Population: the US .gov/.mil registrant set (federal, state, local, tribal and territorial). Each domain was fetched once with a single identified GET request; no crawling. 9,324 of 16,314 were reachable and measurable — the rest were dead, parked, or bot-walled and are excluded rather than counted as failures.
• CMS and version are inferred from response fingerprints and the generator meta. A large Unknown share is expected in government (bespoke and vendor platforms) and is excluded from per-platform cuts, so counts under-report rather than over-report.
• Signals are presence checks, not quality audits. “has structured data” means schema was detected, not that it was complete or valid.
• End-of-life percentages are of the Drupal sites that expose a version; the true count is a floor.

Sources and related work

The public registries, federal policy, platform lifecycle documentation and research this study builds on, and independent surveys you can check our numbers against.

1. CISA / GSA. .gov registry data (get.gov). The authoritative public list of .gov domains — the population frame for this study. Machine-readable mirror at github.com/cisagov/dotgov-data.
2. Office of Management and Budget. M-23-22: Delivering a Digital-First Public Experience, 2023. Federal policy requiring agencies to modernize public-facing websites — the mandate these signals measure progress against.
3. Digital.gov (GSA). 21st Century Integrated Digital Experience Act (IDEA) — resources.
4. Drupal Association. Drupal 7 end of life. Official EOL status for the largest end-of-life version bucket in this study.
5. Drupal Association. Drupal core release schedule and support policy. The supported-vs-end-of-life boundary (Drupal 7/8/9 vs 10/11) used in our version analysis; independently tracked at endoflife.date/drupal.
6. W3Techs. Usage statistics of content management systems. Independent web-wide platform shares to compare the .gov platform mix against.
7. Le Pochat, V., et al.. Tranco: A Research-Oriented Top Sites Ranking Hardened Against Manipulation, NDSS 2019. The sampling frame for the general-web baseline used throughout.
8. Koster, M., Illyes, G., Zeller, H., Sassman, L.. Robots Exclusion Protocol (RFC 9309), IETF, 2022.

Cite this

Replatform Radar, The State of Government CMS Readiness 2026, July 2026. replatformradar.com/research/government-cms-readiness-2026. Licensed CC BY 4.0 — the dataset and every number on this page may be reused with attribution. See also our higher-education study and the web-wide readiness study. Archived on Zenodo: doi:10.5281/zenodo.21615307. Ready-to-paste formats:

Replatform Radar, "The State of Government CMS Readiness 2026," July 2026, https://replatformradar.com/research/government-cms-readiness-2026. doi:10.5281/zenodo.21615307. Open dataset, CC BY 4.0.

{{cite web |author=Replatform Radar |title=The State of Government CMS Readiness 2026 |date=2026-07-23 |url=https://replatformradar.com/research/government-cms-readiness-2026 |publisher=Replatform Radar |doi=10.5281/zenodo.21615307 |access-date=}}

@misc{replatformradar2026gov,
  author       = {{Replatform Radar}},
  title        = {The State of Government CMS Readiness 2026},
  year         = {2026},
  doi          = {10.5281/zenodo.21615307},
  howpublished = {\url{https://replatformradar.com/research/government-cms-readiness-2026}},
  note         = {Open dataset, licensed CC BY 4.0}
}