.edu, measured on 99 sites
.edu is the top-level domain reserved for accredited US educational institutions, so the suffix is a claim about who runs a site rather than where it sits. There are 99 .edu sites in this index, 1.9% of the 5,294 measured. The figure that separates them from the other domain groups here is not the amount of third-party contact, which is high. It is the near-absence of the thing that is supposed to sit in front of it: 14 of the 99 carried a consent tool the crawl could recognise.
Fourteen banners across ninety-nine sites
Of the 99 .edu sites, 14 carried a recognised consent tool. Four products were seen across them: OneTrust on 5 sites, CookieYes on 5, Osano on 3 and Klaro on 1. None of the twelve example sites listed for this group had one. mit.edu, harvard.edu, stanford.edu, cornell.edu, berkeley.edu, washington.edu, academia.edu, princeton.edu, umich.edu, columbia.edu, yale.edu and upenn.edu were all recorded with no recognised banner. That is a selection of the highest-ranked .edu domains rather than a random draw, but it is a consistent one: the twelve best-known names in the group all sit in the 85 sites without a tool.
Seven companies on the median .edu page
The median .edu site contacts 7 separate companies, which ranks second of the nine domain groups profiled here, and carries 2 tracker-classified companies, also second of nine. Median requests is 30, fourth of nine. So this is not a quiet corner of the web by this index's standards; it is one of the busier ones. The examples run well below and well above that median. umich.edu and upenn.edu contacted 1 company each, mit.edu and princeton.edu 2, stanford.edu and berkeley.edu 3, and harvard.edu, cornell.edu, columbia.edu and yale.edu 5 apiece. academia.edu reached 7 and washington.edu 17, the highest of the twelve.
The second-latest first contact of the nine groups
Median first contact across .edu sites is 1,144 ms, which ranks second of the nine domain groups here, and the quickest single first contact recorded in the group was 54 ms. The examples cover almost the whole plausible range. upenn.edu reached outside at 270 ms and harvard.edu at 272 ms; academia.edu at 584 ms, yale.edu 631 ms, umich.edu 662 ms, berkeley.edu 836 ms, mit.edu 907 ms, princeton.edu 1,021 ms, cornell.edu 1,184 ms, columbia.edu 1,280 ms and stanford.edu 1,592 ms. washington.edu sits far outside the rest at 5,547 ms, with 17 companies contacted over the page load and none of them inside the early window.
Early contact on 89 of 99
On 89 of the 99 sites at least one company was contacted before the banner could be answered, a share of 89.9%. That ranks eighth of the nine domain groups here, the second-smallest share in the set. The median count inside that window is 4 companies, and the median after the window closes is 2, so a meaningful share of a .edu page's outside contact is counted late as well as early. Ten of the twelve examples show every company arriving in the early window, including mit.edu at 2 of 2, harvard.edu 5 of 5, stanford.edu 3 of 3, berkeley.edu 3 of 3, academia.edu 7 of 7, princeton.edu 2 of 2, umich.edu 1 of 1, columbia.edu 5 of 5, yale.edu 5 of 5 and upenn.edu 1 of 1. cornell.edu recorded 3 of its 5 early, and washington.edu none of its 17.
What the absence of a banner does not mean
The two headline numbers on this page pull against each other. A .edu page carries about as much third-party surface as the commercial groups in this index, 7 companies and 2 classified as tracking on the median site, while 85 of the 99 showed no recognised consent tool at all. The crawl records what was present, not why. It does not know whether a site decided it had no banner obligation, uses a product the crawl does not recognise, or serves its notice somewhere the crawl did not reach. What it does establish is that on this sample, an .edu address is a poor predictor of a light page.
| Site | Companies | Before consent | First contact |
|---|---|---|---|
| mit.edu | 2 | 2 | 907 ms |
| harvard.edu | 5 | 5 | 272 ms |
| stanford.edu | 3 | 3 | 1592 ms |
| cornell.edu | 5 | 3 | 1184 ms |
| berkeley.edu | 3 | 3 | 836 ms |
| washington.edu | 17 | 0 | 5547 ms |
| academia.edu | 7 | 7 | 584 ms |
| princeton.edu | 2 | 2 | 1021 ms |
| umich.edu | 1 | 1 | 662 ms |
| columbia.edu | 5 | 5 | 1280 ms |
| yale.edu | 5 | 5 | 631 ms |
| upenn.edu | 1 | 1 | 270 ms |
| Tool | Sites |
|---|---|
| OneTrust | 5 |
| CookieYes | 5 |
| Osano | 3 |
| Klaro | 1 |
Questions
Does a .edu address mean a site is non-commercial?
The registry restricts .edu to accredited US institutions, but nothing in these measurements distinguishes commercial from non-commercial behaviour. The median .edu site here contacts 7 companies and carries 2 classified as tracking, which ranks second of the nine domain groups profiled on both counts.
Which consent products appear on .edu sites?
Four were recognised across the 14 sites carrying a tool: OneTrust on 5 sites, CookieYes on 5, Osano on 3 and Klaro on 1. These are counts of sites where each product was seen alongside the others in the group, not shares of a single total.
Why do so few of these sites show a banner?
The crawl does not record reasons, only presence. It found a recognised tool on 14 of 99 sites. It cannot tell you whether the remaining 85 made a deliberate choice, use something it does not recognise, or show a notice on a page it did not visit.
What was the slowest first contact in the group?
washington.edu, at 5,547 ms, well beyond the group's median of 1,144 ms. It contacted 17 companies over the full page load, more than any other example here, and none of them fell inside the early window.