siteimproveanalytics.com in the Consent Gap Index

A public-sector cohort

Ten of the twelve entries in this entity's named sample are universities or US government bodies: cornell.edu, berkeley.edu, yale.edu, uci.edu, uchicago.edu, jhu.edu, noaa.gov, sec.gov, bls.gov and hhs.gov, with apa.org and xerox.com the outliers. That is not a coincidence of sampling; it is what this product's customer base looks like. Its counterpart host on the .io domain, part of the same service, appears on 53 of the 58 sites, and a US government shared web-services host on 15 of them (25.9 percent). Institutions that buy accessibility auditing are institutions with an obligation to be auditable, which makes their third-party behaviour worth measuring separately from a commercial site's.

First contact at 996 ms, third-latest of the 88

First contact on the median site here landed at 996 milliseconds — the 3rd-latest median of the 88 companies in this index. The fastest first contact recorded on any of these sites was 257 milliseconds. A page that waits most of a second before reaching any third party is a page whose third-party code is not in the critical path. That is the ordinary consequence of a slower publishing stack and a lighter commercial motive, and it is the single most distinctive number on this page.

Light pages by every count

The median site contacts 7 distinct companies, 2 of them classified as trackers, over 26 requests to third-party hosts. That request count is the 8th-lowest of the 88 companies indexed, and the count of companies contacted before an answer could be registered — 5 on the median site — ranks 57th of 88, with only nine companies here sitting lower and 22 more reporting the same 5. Two more companies were contacted after the consent window closed. By every measure of volume this is a restrained corner of the web, which makes the next figure the awkward one.

Nine banners, and Google on 55 of 58

A consent tool was recognised on just 9 of the 58 sites. Four were seen alongside them: OneTrust on 6, Klaro on 1, Osano on 1 and Cookiebot on 1. Meanwhile early contact happened on 55 of the 58 sites, 94.8 percent, and Google was present on 55 of the 58 as well (94.8 percent). Cloudflare appears on 14 sites (24.1 percent), Font Awesome and jsDelivr on 10 each (17.2), Adobe Fonts on 9 (15.5), Crazy Egg, Meta and Microsoft Clarity on 8 each (13.8). Restrained in volume does not mean absent in kind: the pages are light, they start late, and almost all of them still reach Google.

The named sites, one by one

yale.edu contacted 5 companies with all 5 early, at 631 ms; sec.gov 5 of 5 at 663 ms; bls.gov 5 of 5 at 1,064 ms; berkeley.edu 3 of 3 at 836 ms. Others defer most of it. jhu.edu contacted 6 companies with 2 early at 1,878 ms, xerox.com 9 with 3 at 1,697 ms, noaa.gov 8 with 4 at 1,341 ms, hhs.gov 7 with 4 at 1,377 ms, uchicago.edu 9 with 5 at 1,401 ms, uci.edu 5 with 3 at 1,160 ms and cornell.edu 5 with 3 at 1,184 ms. apa.org is the only site in the sample with a banner recorded and zero companies contacted early, with first contact at 2,279 ms.

Why this group is worth its own page

Most discussion of third-party tracking is a discussion about commercial publishing, because that is where the numbers are largest. This entity covers the opposite case: 58 sites belonging mostly to institutions that people cannot choose to avoid, carrying a light and slow-starting stack, showing a banner on 9 of them, and reaching Google on 55. Neither the volume figures nor the timing figures on their own would tell you that.

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.

• GoogleTogether on 55 of these sites
• siteimproveanalytics.ioTogether on 53 of these sites
• digitalgov.govTogether on 15 of these sites
• CloudflareTogether on 14 of these sites
• jsDelivrTogether on 10 of these sites
• Font AwesomeTogether on 10 of these sites
• Adobe FontsTogether on 9 of these sites
• Crazy EggTogether on 8 of these sites