siteimproveanalytics.io: universities, agencies, few banners

A public-sector and academic population

Of the twelve named sites recorded for this hostname, six are universities and four are United States federal agencies: cornell.edu, berkeley.edu, yale.edu, uci.edu, uchicago.edu and jhu.edu on one side, noaa.gov, sec.gov, bls.gov and hhs.gov on the other, with apa.org and xerox.com making up the rest. A second hostname under the same name was recorded alongside it on all 53 sites, Google on 94.3 percent, and a United States federal web-services host on 28.3 percent. That last figure only reinforces the pattern.

Seven banners across 53 sites

A recognisable consent tool was detected on seven of the 53 sites, 13.2 percent. OneTrust was seen alongside four of them, and Klaro, Osano and Cookiebot alongside one each. Seven sites out of 53 is the figure worth carrying away from this page: an institutional population running an analytics product, on sites that overwhelmingly do not present a cookie dialogue the crawl could recognise. Sites can carry more than one banner, so those four counts are what was seen alongside, not a division of the seven.

Late first contact, and still early by the measure

The median site here made its first third-party request at 850 milliseconds, which ranks twelfth of the 88 companies in this index when ordered from latest to earliest, so these pages are slower to reach outward than most. The quickest observed was 257 milliseconds. Even so, 52 of the 53 sites, 98.1 percent, had at least one company contacted before the point a consent choice could be made, a share ranking eleventh of 88. Only one site had none.

Quiet pages that keep loading

The median site reached seven separate companies, ranking 66th of 88, made 26 requests, ranking 82nd, and carried one host classified as tracking, ranking 82nd. These are light pages. But the median count of companies reached inside the early window is five, and the median count first seen after the window closes is two. A small third-party set here is not a set that has finished by the time the banner is dealt with. The two figures sit oddly together and the corpus does not reconcile them for you. A page can hold a short list of third parties and still be reaching new ones a second or two in, particularly when its own first request is late, and on these sites the median first request is late.

The individual sites

berkeley.edu reached three companies, all three inside the early window, at 836 milliseconds. sec.gov reached five, all early, at 663 milliseconds, and bls.gov five, all early, at 1,064 milliseconds. yale.edu reached five, all early, at 631 milliseconds. Elsewhere the split is wider: cornell.edu five with three early at 1,184 milliseconds, uci.edu five with three early, uchicago.edu nine with five early, hhs.gov seven with four early, noaa.gov eight with four early at 1,341 milliseconds, jhu.edu six with two early at 1,878 milliseconds, xerox.com nine with three early at 1,697 milliseconds. apa.org, one of the seven sites with a detected banner, reached six companies with none inside the window and first contact at 2,279 milliseconds.

Seen on the same pages

Companies most often contacted by the same sites. Co-presence only — the crawl records what arrived, not who called whom.

• siteimproveanalytics.comTogether on 53 of these sites
• GoogleTogether on 50 of these sites
• digitalgov.govTogether on 15 of these sites
• CloudflareTogether on 13 of these sites
• jsDelivrTogether on 9 of these sites
• Font AwesomeTogether on 9 of these sites
• Adobe FontsTogether on 8 of these sites
• MetaTogether on 7 of these sites