Drupal Powers 33% of Live U.S. Federal Websites in New CMS Census
Drupal was detected on 167 of 503 accessible, live U.S. federal government websites, or 33.2%, in a new census by digital agency Gspikes. Published on 11 November 2026 by founder and CEO Shaker Abady, the study examined the content management systems serving websites associated with domains in the official U.S. .gov registry. The finding describes the sites researchers could access and identify, not Drupal's verified share of every federal government website.
Across all levels of government, Gspikes identified 8,827 live websites from 16,999 registered domains. It detected Drupal on 433 of them (4.9%), compared with WordPress on 2,433 (27.6%) and CivicPlus on 1,991 (22.6%). Drupal was the most frequently identified CMS among the 503 live federal websites, although approximately 40% of that federal group had no detectable CMS fingerprint. The study therefore cannot establish which platforms power all of those unidentified sites.
Drupal's Presence Varies by Level of Government
Gspikes detected Drupal on 65 of 440 live state and territory websites (14.8%), 51 of 1,408 county websites (3.6%) and 136 of 5,641 city, town and village websites (2.4%). WordPress led among state and territory sites, while WordPress and commercial government website vendors accounted for substantially larger shares of local government websites. These figures describe platforms detected on the sampled homepages, rather than all systems used by each government organisation.
The study also identified Drupal version differences within the 433 sites. Gspikes classified 83 as running Drupal 7, 8 or 9, versions that had passed end of life by the census date. That includes 55 of the 136 city Drupal sites (40%) but six of the 167 federal Drupal sites (about 4%). It identified 165 more sites as running Drupal 10. Under the Drupal core release schedule, Drupal 10 reaches end of life on 9 December 2026. A version fingerprint does not by itself establish a site's full patch status or security configuration.
What the Census Could and Could Not Reach
Gspikes used the CISA .gov registry updated on 1 November 2026 and requested homepages on 2 November from a server in Amman, Jordan. The automated checks tried HTTPS and, where appropriate, alternate hosts and HTTP. CMS identification used publicly visible response headers, markup, scripts and platform-specific markers. The researchers describe two rounds of manual spot checks and revisions to their detection rules after finding false matches and non-government placeholder pages.
A further 3,194 registered domains blocked or challenged the requests, including sites belonging to several major federal departments. Such domains were not assigned guessed CMS platforms. Other domains redirected, served placeholder pages, returned errors or did not respond. Content delivery networks, decoupled frontends and other configurations can also obscure CMS fingerprints. For these reasons, the census is a measured snapshot of accessible sites rather than a comprehensive inventory of all U.S. government technology.
The findings do not measure website accessibility compliance. The term accessible in this context refers to whether the researcher's automated request could reach and inspect a live homepage, not whether the site meets accessibility standards. The researchers also caution that a homepage may not represent the platform used by every application or subsite operated by the same agency.
Why the September and November Counts Differ
The DropTimes previously covered a September 2026 Gspikes study that identified Drupal on 424 responding federal, state and territory domains. To compare methods, Gspikes recalculated its November federal-and-state results using the earlier domain-counting approach and found 413 Drupal detections among 1,518 responding domains. That figure includes 181 domains that forward to another government website running Drupal. Counting distinct live websites instead produces 232 Drupal sites among 943 federal and state sites (24.6%). Differences in redirects, coverage and counting rules mean the two studies should not be read as evidence of a sudden change in Drupal adoption.
The census also provides a different perspective from The DropTimes' analysis of Drupal's wider CMS market share. That analysis drew on W3Techs data about websites with an identified CMS, whereas Gspikes examined a defined set of U.S. government domains using its own detection methodology. The much larger measured Drupal presence among the federal sites available to Gspikes does not directly contradict its lower share across the wider web. Gspikes makes its site-level CSV dataset available for further examination; The DropTimes has not independently rerun the full census.
