X across 57 measured sites

Where 80.7 percent sits among 88 companies

46 of the 57 sites had at least one company contacted before a consent choice could have been recorded. That is 80.7 percent, and it ranks 87 of the 88 companies in this index: eighty-six entries here sit on populations where early contact is more common.

Eleven sites had none at all. In an index where most entries sit in the nineties, eleven out of fifty-seven is a lot, and it is worth understanding before reading anything else on this page.

Two clusters, and a median that describes neither

Median first contact across these sites was 760 ms, rank 27 of 88 where rank 1 is the longest wait. But almost none of the individual sites are anywhere near 760 ms.

One cluster starts immediately and holds nothing back. udemy.com contacted 13 companies with 12 of them inside the window, first contact 175 ms, behind a OneTrust banner. testbook.com: 11 companies, all 11 early, at 248 ms. plos.org: 9 and 9, at 269 ms. heylink.me: 11 and 11, at 400 ms. split.io: 14 companies, 11 early, at 535 ms.

The other cluster idles for seconds and contacts nothing inside the window. nypost.com: 8 companies, none early, first contact 2882 ms, behind a OneTrust banner. alphapolis.co.jp: 12 companies, none early, at 2642 ms. suumo.jp: 10 companies, none early, at 2162 ms. jhu.edu sits between them at 1878 ms with 2 of 6 early.

A median of 760 ms is arithmetically correct and describes no site in either group. On a set shaped like this, the medians are a summary of the distribution and not a portrait of a typical page.

The pages are mid-weight but tracker-dense

The median site here contacted 9 separate companies across 49 requests — rank 46 and rank 39 among the 88 companies, squarely mid-field. But 4 of those companies were classified as tracking, which ranks 18 of 88, and the median count of companies arriving before the window closes is 6, rank 37 of 88, level with 19 others.

So the pages carrying X's hosts are not especially large. They are pages where a comparatively high proportion of what does load is catalogued as tracking, and where most of it lands early when it lands early at all.

Only 12 of 57 showed a banner

12 of the 57 sites carried a consent tool the crawl could detect: OneTrust seen alongside on 6, an IAB TCF consent tool on 3, CookieYes on 2, Cookiebot on 1. A site can run more than one and each recognised tool is counted, so these are not slices of the 12.

On the other 45 there was nothing to answer. That makes the low early-contact share more interesting rather than less: these are mostly sites with no banner, and they still land at rank 87 of 88 for early contact, because a good number of them simply load slowly and late.

X was present on the 46 sites where early contact happened; the crawl counts that per site and does not record which companies arrived inside the window.

It almost never appears without Google

Google was present on 56 of the 57 sites — 98.2 percent, the top of its co-presence list. After that the list is other social and analytics pixels: Meta Pixel on 19 sites, 33.3 percent; Meta on 17; LinkedIn on 15, 26.3 percent; TikTok Pixel on 11 and tiktokw.us on 9; Microsoft and Microsoft Clarity on 10 each. jsDelivr and yimg.jp appear on 8 each.

That is the profile of a site that runs several social advertising tags at once rather than one. One site in three in this set carried Meta Pixel as well, and one in four carried LinkedIn. Whether any given page looks like that is a question about that page, and this entry can only report the rates across these 57.

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 56 of these sites
• Meta PixelTogether on 19 of these sites
• MetaTogether on 17 of these sites
• LinkedInTogether on 15 of these sites
• TikTok PixelTogether on 11 of these sites
• MicrosoftTogether on 10 of these sites
• Microsoft ClarityTogether on 10 of these sites
• jsDelivrTogether on 8 of these sites