Original research · July 2026

Nobody Voted for This Template

The most consequential web-standards decision-maker in American local government is a vendor no voter has heard of — and when its template omits schema, a quarter of local government goes dark at once.

Nobody chose this

Thousands of American towns, counties and special districts each bought a website. Each purchase was a local decision — a council vote, a procurement cycle, a budget line. Collectively, without anyone deciding it, they bought one template. A single specialist platform is now the front door to 20.7% of the live US government web — 2,499 sites — and the three largest platforms together account for 47.1%.

That concentration is not, in itself, a scandal. It is an amplifier. One good default in a template like that protects millions of residents at once. One omission erases them just as fast — and nobody votes on either.

The natural experiment

Two fleets of almost exactly the same size make the point without any modelling. The largest government-specialist platform ships one template to 2,499 governments. WordPress is an open platform on which 2,458 governments each made their own independent choices. Same sector, same kinds of towns, same budgets, same year.

The darkness is not a front-door effect. We fetched one interior page per site wherever a content link was available: across 2,120 Specialist Vendor A interior pages, 0.2% carried structured data. The fleet is dark all the way through.

Hiring the specialist is the risk factor

You would assume that hiring the company which specializes in government websites produces a better civic web than a volunteer wiring up WordPress. For the layer that now decides whether a resident’s question gets an authoritative answer, the opposite is true — in both sectors we measured.

The market, vendor by vendor

Every platform running at least 25 identifiable US government sites, largest fleet first. A vendor is only assigned on a platform-specific marker, so these shares are conservative: we identified a platform on 67.8% of reachable government sites and 78.6% of university sites.

A note on names. This public edition identifies government- and education-specialist platforms by neutral label (Specialist Vendor A, B, C…) rather than by company name. The subject here is a category-level pattern — what happens when one template serves thousands of governments — not the conduct of any individual company. Labels are stable and ordered by fleet size, so every figure stays internally consistent, and the method is published in full so anyone can recompute the same measurements independently. General-purpose platforms are named: they are the comparison group and are largely open-source projects rather than vendors under scrutiny.

The fix is one template change

This is the unusual part. Nothing here requires 2,499 towns to do anything. A single specialist platform adding GovernmentOrganization and LocalBusiness markup to its base template would move a fifth of the American government web in one deploy — the largest single improvement in civic machine-readability available to anyone, and it sits with a handful of product teams.

That is also why this study reports platform categories rather than naming towns or companies. The towns did not make this choice, and the finding is about what template concentration does — not about the conduct of any individual vendor.

Method and limitations

• Population: the US .gov/.mil and .edu registrant sets — 18,682 domains, of which 14,057 (75.2%) were reachable. Each domain was fetched once at the homepage with a single identified request, plus at most one interior page. No crawling.
• Vendor is inferred from response fingerprints (CDN and script hosts, cookies, headers, generator meta, footer markers) and assigned only on a platform-specific marker, so counts under-report rather than over-report. Embedded third-party services — meeting portals, translation and accessibility widgets, mapping — are recorded separately and never counted as the site’s vendor.
• “Machine-readable” means schema.org structured data was detected as JSON-LD. It is a presence check, not a quality audit: we do not assess whether the markup is complete, valid, or well-chosen.
• Interior-page results cover the subset of sites where a content link was available from the homepage (2,591 specialist and 1,323 general-purpose government pages).
• Readiness scores are joined from the single-fetch scans behind our government and higher-education studies, and carry those studies’ limitations.

Sources and related work

The public registries, standards and policy this study builds on.

1. CISA / GSA. .gov registry data (get.gov). The authoritative public list of .gov domains — the population frame for the government half of this study.
2. Schema.org. Schema.org — shared structured-data vocabulary. The vocabulary whose presence we measure; GovernmentOrganization and LocalBusiness are the types most relevant to a local government homepage.
3. Google Search Central. Introduction to structured data markup. Why machine-readable markup determines whether a page can be quoted rather than merely found.
4. Office of Management and Budget. M-23-22: Delivering a Digital-First Public Experience, 2023. Federal policy on modernizing public-facing government websites.
5. Digital.gov (GSA). 21st Century Integrated Digital Experience Act (IDEA) — resources.
6. Koster, M., Illyes, G., Zeller, H., Sassman, L.. Robots Exclusion Protocol (RFC 9309), IETF, 2022. The crawl-permission standard this study's single-fetch method respects.
7. W3Techs. Usage statistics of content management systems. Independent web-wide platform shares, for comparison against the sector mix reported here.

Cite this

Replatform Radar, Nobody Voted for This Template, July 2026. replatformradar.com/research/civic-monoculture-2026. Licensed CC BY 4.0 — the dataset and every number on this page may be reused with attribution. Ready-to-paste formats:

Replatform Radar, "Nobody Voted for This Template," July 2026, https://replatformradar.com/research/civic-monoculture-2026. Open dataset, CC BY 4.0.

{{cite web |author=Replatform Radar |title=Nobody Voted for This Template |date=2026-07-28 |url=https://replatformradar.com/research/civic-monoculture-2026 |publisher=Replatform Radar |access-date=}}

@misc{replatformradar2026monoculture,
  author       = {{Replatform Radar}},
  title        = {Nobody Voted for This Template},
  year         = {2026},
  howpublished = {\url{https://replatformradar.com/research/civic-monoculture-2026}},
  note         = {Open dataset, licensed CC BY 4.0}
}