.net
The .net suffix was set up for the companies that run network infrastructure — access providers, registries, backbone operators — and it never entirely lost that character after registration opened to everyone. A lot of what carries a .net address today is not a website in the ordinary sense but a piece of plumbing: a delivery endpoint, a messaging backend, an ad server. The rest is the ordinary web, often a business whose .com was already taken. This index measured 245 .net sites, 4.6% of the 5,294 pages in it, which makes .net the third-largest of the nine top-level domain groups broken out here.
Lighter than most groups, but not the lightest
The median .net page in this sample reaches 4 outside companies over 20 third-party requests, with 1 of those companies classified as tracking. Against the other eight top-level domains measured here, that median company count ranks seventh of nine and that request count sixth of nine — the lower half of the table on both, without being the floor of it.
That is roughly what the mix of registrants would predict. A delivery endpoint or a router's remote-access page has nothing to sell and nothing to measure, so it loads almost nothing. A consumer portal on the same suffix behaves like any other consumer portal. The median falls between the two, and it describes neither.
228 of the 245 had contact before the banner could be answered
On 228 of the 245 .net sites — 93.1% — at least one outside company was contacted before a consent choice could have been made. That share ranks fourth of the nine top-level domains here, so it is high but not the highest of them.
The median site made its first third-party contact 649 ms into the load, the sixth-longest median of the nine groups. The earliest anywhere in the .net sample was 75 ms, which is fast enough that no banner could have rendered, let alone been read.
Two populations under one suffix
The sample splits cleanly when you look at individual sites. On the infrastructure side: whatsapp.net reached 1 company, doubleclick.net 1, myfritz.net 2 with first contact at 388 ms, yandex.net 3 with all 3 inside the window at 222 ms, b-cdn.net 3 with all 3 inside it. Slightly larger service brands sit just above: cloudflare.net at 4 companies behind a OneTrust banner, gandi.net at 3, fastly.net at 7.
On the consumer side the numbers change scale. comcast.net reached 21 companies, more than five times the group median, with 1 of them inside the window and first contact at 1,966 ms. researchgate.net reached 5, all 5 inside the window, behind an IAB TCF consent tool. cpanel.net reached 6, all 6 inside it. netflix.net is the sample's counter-example: 6 companies, none of them inside the window, first contact at 2,136 ms behind a OneTrust banner.
78 of the 245 showed a banner at all
A consent tool was detected on 78 .net sites. Within that group an IAB TCF consent tool was seen on 46 and OneTrust on 23, with Osano on 3, Cookiebot on 3, Usercentrics on 2 and CookieYes on 1. These are counts of sites where each tool was seen alongside the others, not slices of a whole — a single site can carry more than one banner, so they cannot be read as shares.
The other 167 sites showed no detected banner. For much of this suffix that is unremarkable: a hostname that exists to serve files to other machines has no visitor to ask.
For the median page, the load is already decided
After the consent window closes, the median .net site adds no further companies at all — the median number arriving after the window is 0. Whatever a visitor to a typical page on this suffix was going to be exposed to had already happened by the time an answer was possible.
That is a different statement from the site being heavy or light. A page with 4 companies and a page with 21 can both finish their third-party contact early; what changes between them is how much there is to finish.
| Site | Companies | Before consent | First contact |
|---|---|---|---|
| whatsapp.net | 1 | 1 | 1039 ms |
| fastly.net | 7 | 3 | 670 ms |
| doubleclick.net | 1 | 1 | 955 ms |
| gandi.net | 3 | 2 | 1686 ms |
| cloudflare.net | 4 | 2 | 343 ms |
| myfritz.net | 2 | 2 | 388 ms |
| yandex.net | 3 | 3 | 222 ms |
| b-cdn.net | 3 | 3 | 574 ms |
| netflix.net | 6 | 0 | 2136 ms |
| cpanel.net | 6 | 6 | 1470 ms |
| researchgate.net | 5 | 5 | 415 ms |
| comcast.net | 21 | 1 | 1966 ms |
| Tool | Sites |
|---|---|
| IAB TCF consent tool | 46 |
| OneTrust | 23 |
| Osano | 3 |
| Cookiebot | 3 |
| Usercentrics | 2 |
| CookieYes | 1 |
Questions
Is .net still limited to network companies?
No. Registration has been open for a long time, and the index shows both kinds of registrant side by side: infrastructure hostnames such as b-cdn.net and myfritz.net that reach two or three outside companies, and ordinary consumer sites such as comcast.net that reach twenty-one.
How many .net sites are in this index?
245, out of 5,294 pages measured in total — 4.6% of the corpus, and the third-largest of the nine top-level domain groups reported here.
What was the earliest first contact on a .net site?
75 ms after the page began loading. The median across the 245 sites was 649 ms, which ranks sixth-longest of the nine top-level domains measured here.
Do lighter pages make their first contact later?
Not in this sample. .net ranks seventh of nine on median company count but sixth of nine on median time to first contact, so being lighter than most groups did not make it slower to reach the first outside company. The number of third parties and the timing of the first one are set by different decisions.
Why do so few .net sites show a consent banner?
78 of the 245 did. Whether a site shows one depends on where its visitors are and what kind of business it runs, and a large share of this suffix is machine-facing infrastructure with no visitor-facing page to put a banner on.