visualwebsiteoptimizer.com

A script that has to run before the page is drawn

This is the part of the measurement worth understanding before reading any of the numbers. A tool that swaps out a headline has to make its decision before the browser paints, or the visitor sees the original headline and then watches it change — the flicker that split-testing vendors spend most of their documentation explaining how to avoid. The standard fix is to load the script as early in the document as possible and let it block rendering until it has chosen.

So pages running a tool of this shape have a structural reason for reaching an outside company in the first moments of the load, quite separate from any advertising. That is a general statement about how such tools are installed, not a claim about which company this index saw arriving first on any particular page.

77 of 80 had contact before the banner could be answered

Early contact occurred on 77 of the 80 sites, 96.3%, which ranks 21st of the 88 companies in this index — the upper quarter of that table. On the median site, 6 companies had been reached inside the consent window.

The median first contact landed at 747 ms, ranking 30th of 88, and the earliest anywhere in this group was 128 ms. Once the window closes, the median site adds no further companies at all.

Not an advertising profile

The pages themselves are unremarkable in weight and quiet in composition. The median site reaches 7 outside companies over 41 third-party requests, ranking 62nd and 51st of the 88 companies here. Of those 7, the median number classified as tracking is 1 — the 75th placement of 88, near the bottom of that table.

That combination is the interesting one: a high early-contact share on pages that are not carrying an advertising stack. Whatever is producing the early contact on these sites, it is not a page full of tracking vendors, because by the index's own classification these pages barely have any.

Reference works, standards bodies, developer tools

Nine of the twelve example sites published for this group had every outside company they contacted fall inside the consent window. merriam-webster.com reached 12 companies with all 12 inside it. britannica.com reached 9 with all 9, behind an IAB TCF consent tool. weforum.org and oecd.org each reached 6 with all 6 inside the window, cpanel.net 6 with all 6, checkpoint.com 6 with all 6 at 222 ms, branch.io 6 with all 6, braze.com 5 with all 5 and taboola.com 5 with all 5.

The three exceptions are the slower loaders. qualtrics.com reached 8 with 6 inside the window at 1,085 ms, segment.io 12 with 5 inside it at 1,668 ms, and plesk.com 12 with only 2 inside it at 966 ms. In each of those, most of the page's third-party contact happened after the window rather than inside it.

Seven sites in ten show a banner

A consent tool was detected on 56 of the 80 sites, 70% of the group. OneTrust was seen on 33 of them, CookieYes on 8, and an IAB TCF consent tool, Cookiebot and Osano on 4 each, with Usercentrics on 2. Those are counts of sites where each tool was seen alongside the others rather than a division of the 56, since a site can carry more than one banner.

The co-presence list points the same way. Google is on 82.5% of the sites and OneTrust on 40%, with Google Tag Manager and Cloudflare Analytics on 13.8% each, Microsoft Ads on 12.5% and LinkedIn on 10%. A company that pays for controlled experiments on its own funnel is generally a company large enough to have bought a compliance product too.

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 66 of these sites
• OneTrustTogether on 32 of these sites
• Cloudflare AnalyticsTogether on 11 of these sites
• Google Tag ManagerTogether on 11 of these sites
• Microsoft AdsTogether on 10 of these sites
• CloudflareTogether on 9 of these sites
• LinkedInTogether on 8 of these sites
• cdn-cookieyes.comTogether on 8 of these sites