What the .io sites in this index actually load
.io is a country-code top-level domain, assigned to the British Indian Ocean Territory and then adopted almost entirely by people who read it as input/output. Developer tools, APIs, SDK vendors and documentation sites live there. This index measured 104 of them, 2 percent of the 5,294 sites in the crawl, and among the nine top-level domains recorded here .io is the one where the early-contact share reached 100 percent: on all 104 sites, at least one company was contacted before a consent choice could have been recorded.
104 out of 104
The 100 percent figure is rank 1 of the nine top-level domains in this index, and it is the maximum for that measure. So is the median count of companies contacted inside that window: 5, also rank 1 of 9.
That second number is the more surprising of the two. It is easy to reach 100 percent on a measure like this if the bar is one stray request. Five companies at the median is not a stray request. It means the typical .io site in this crawl had most of a third-party stack open before there was anything for a visitor to answer.
These are not heavy pages
The median .io site contacted 5 separate companies across 24 requests, with 1 of them classified as tracking. Among the nine top-level domains here that ranks 5th on companies, 5th on requests and 7th on trackers — the middle of the field or below it.
So the 100 percent is not a story about bloat. github.io is the clearest case in the set: 1 company contacted on the whole page, and that single company arrived before the window closed, at 1253 ms. ipinfo.io reached 3 companies, all 3 early, first contact 201 ms. readthedocs.io: 4 and 4, at 563 ms. sentry.io: 4 and 4, at 515 ms. A light page and an early page are separate properties, and .io is the domain where that comes apart most visibly.
Two thirds of them never asked anything
Only 38 of the 104 sites carried a consent tool the crawl could detect. On the other 66 there was no banner to answer, which is one honest explanation for why the early-contact share is what it is: on most of these sites nothing was ever put to the visitor.
Among the 38, OneTrust was seen alongside on 14 sites, an IAB TCF consent tool on 7, Cookiebot on 5, Osano on 5, CookieYes on 3 and Usercentrics on 2. Those counts are not slices of the 38 — a site can carry more than one banner, and the crawl counts each one it recognises — so do not read them as a breakdown that ought to add up.
Where the banners are, the pages are bigger
The banner-carrying examples in this set are the busier ones. adobe.io contacted 8 companies with 6 of them early, first contact 1024 ms, behind a OneTrust banner. intercom.io: 7 companies, 5 early, 798 ms, also OneTrust. wpguardian.io: 7 companies, all 7 early, first contact 1521 ms, behind a Usercentrics banner. branch.io: 6 and 6, at 807 ms, OneTrust. oxylabs.io contacted 12 with only 2 early, first contact 857 ms, behind a Cookiebot banner.
And the heaviest .io example carries no detectable banner at all: liftoff.io, 14 companies, all 14 of them inside the window, first contact 755 ms.
Timing
Median first contact on a .io site was 515 ms. Among the nine top-level domains here that ranks 9th of 9, where rank 1 is the longest wait — the short end of that list. The earliest single first contact recorded was 55 ms.
Individual sites still range widely, from ipinfo.io at 201 ms to wpguardian.io at 1521 ms. The median count of companies appearing after the window closed is 0, which fits the rest: on a typical .io page the third-party contact happens at the start and then stops.
| Site | Companies | Before consent | First contact |
|---|---|---|---|
| sentry.io | 4 | 4 | 515 ms |
| github.io | 1 | 1 | 1253 ms |
| adobe.io | 8 | 6 | 1024 ms |
| oxylabs.io | 12 | 2 | 857 ms |
| ipinfo.io | 3 | 3 | 201 ms |
| pushy.io | 3 | 2 | 1358 ms |
| branch.io | 6 | 6 | 807 ms |
| intercom.io | 7 | 5 | 798 ms |
| itch.io | 2 | 2 | 1131 ms |
| wpguardian.io | 7 | 7 | 1521 ms |
| liftoff.io | 14 | 14 | 755 ms |
| readthedocs.io | 4 | 4 | 563 ms |
| Tool | Sites |
|---|---|
| OneTrust | 14 |
| IAB TCF consent tool | 7 |
| Cookiebot | 5 |
| Osano | 5 |
| CookieYes | 3 |
| Usercentrics | 2 |
The six most common are listed. The remaining 2 of 38 sites with a banner carry something outside that six. One tool is recorded per site, so these counts never overlap.
Questions
Why do so many technology companies use .io?
It reads as input/output, it was available when .com equivalents were not, and it became a convention in developer tooling. Formally it is the country-code domain for the British Indian Ocean Territory. This index measured 104 .io sites.
Do .io sites track more than other domains?
Not by volume. The median .io site contacted 5 companies over 24 requests with 1 classified as tracking, which is mid-field among the nine top-level domains here. What stands out is timing, not weight.
What does 100 percent early contact mean here?
On each of the 104 sites, at least one company was contacted before a consent choice could have been recorded. It is a fact about the sites, and the crawl does not record which company arrived first on any of them.
How many .io sites showed a cookie banner?
38 of the 104. On the remaining 66 the crawl detected no consent tool at all, so there was nothing for a visitor to answer in the first place.