What 86 .gov sites do before you can answer anything
.gov is a restricted top-level domain. Only United States government bodies can hold one — federal agencies, states, counties, tribal governments — and the registry checks. That restriction is why people arrive at a page like this expecting the numbers to look different from a commercial site's. In this index, 86 sites carry a .gov address: 1.6% of the 5,294 measured, and the smallest of the nine top-level domains broken out here. They do look different, though not in the way the restriction implies.
One site in 86 had a banner the crawl could detect
Across the whole group, consent software was detected on a single site, running Osano. The other 85 asked nothing.
Early contact is still recorded for them. The pre-consent window is the opening stretch of a page load, and it exists whether or not there is a question on screen; on a site with no banner it simply means nothing was ever going to be answered. On 80 of these 86 sites, at least one company was contacted inside that window — 93%, which ranks fifth of the nine top-level domains here. The median count of companies arriving in the window was 3.
The pages themselves are light
The median .gov site contacted 5 companies over 19 requests, with 1 of those companies classified as tracking against the vendor catalogue. Nineteen requests is a thin page by the standards of this index: it ranks seventh of nine on request volume and sixth on companies contacted.
So the picture is not one of heavy commercial instrumentation. It is a small number of third parties — fonts, video, analytics, the ordinary furniture of a public information site — reached early, on sites that never put a question in front of the visitor.
First contact is slower here than in most of the index
The median .gov site did not reach a third party until 927 ms into the load, which ranks third of nine, where rank 1 is the slowest group. The quickest first contact recorded anywhere in the group was 237 ms, on nasa.gov.
Slowness of this kind is a statement about when, not about whether. A page that waits nearly a second before its first outbound connection is usually one where the third-party code sits below the main content rather than in the head of the document. It still gets there.
Twelve of the eighty-six, by name
nist.gov contacted 5 companies, 4 of them inside the window, first contact at 611 ms. fda.gov: 5 and 4. epa.gov: 5 and 4. whitehouse.gov: 4 companies with 3 inside the window at 690 ms. state.gov: 4 and 4, first contact at 927 ms. ftc.gov: 6 and 6, at 615 ms. nasa.gov: 3 and 3 at 237 ms. cdc.gov reached only 2 companies but did not reach them until 1,711 ms; usda.gov also 2 and 2. ca.gov contacted a single company. noaa.gov is the busiest of the federal examples at 8 companies, 4 of them inside the window, first contact at 1,341 ms.
None of the twelve had a consent banner.
maricopa.gov is the shape none of the others are
The county site maricopa.gov contacted 9 companies, more than any other example here, and none of them inside the window. First contact came at 3,205 ms.
That combination — a heavier page, and nothing at all in the opening window — is what the rest of this group is not. It is a useful reminder that the two numbers on this page measure different things. Company count says how much of the page belongs to someone else. Early contact says how much of it happened before a visitor had any say. A site can score badly on one and cleanly on the other, and on .gov, 80 of 86 do the reverse of maricopa.gov: fewer companies, reached sooner.
| Site | Companies | Before consent | First contact |
|---|---|---|---|
| nist.gov | 5 | 4 | 611 ms |
| cdc.gov | 2 | 2 | 1711 ms |
| nasa.gov | 3 | 3 | 237 ms |
| ca.gov | 1 | 1 | 1053 ms |
| noaa.gov | 8 | 4 | 1341 ms |
| fda.gov | 5 | 4 | 516 ms |
| usda.gov | 2 | 2 | 1327 ms |
| state.gov | 4 | 4 | 927 ms |
| epa.gov | 5 | 4 | 1325 ms |
| whitehouse.gov | 4 | 3 | 690 ms |
| maricopa.gov | 9 | 0 | 3205 ms |
| ftc.gov | 6 | 6 | 615 ms |
| Tool | Sites |
|---|---|
| Osano | 1 |
Questions
Who can register a .gov domain?
United States government organisations only — federal agencies, state and local governments, tribal nations — through the registry that administers the zone. That is what makes the group a coherent thing to measure: every site in it is run by a public body.
Why did only one .gov site show a consent banner?
This index records what was detectable on one cold load from a single location. US federal and state sites are generally not operating under an EU-style consent requirement for their domestic audience, and 85 of the 86 measured here presented no banner at all.
Does early contact mean these sites did something wrong?
No. Early contact has ordinary causes — a font host, a video embed, an analytics beacon placed high in the document. The measurement records that a third party was reached before anything could be answered, and nothing about the lawfulness or intent of that contact.
Are .gov sites lighter than the rest of the index?
Within this group, yes by its own ranks: 19 requests and 5 companies on the median site put it seventh of nine on volume and sixth of nine on companies among the top-level domains broken out here.