Alpine.js, and the 84 sites this index found it on
Alpine.js is a small JavaScript library — a few kilobytes — for adding behaviour to HTML the server has already rendered. Dropdowns, tabs, modals and toggles are declared as x-data and x-on attributes in the markup itself, with no build step and no component tree. Developers reach for it when a full framework would be too much machinery for the job. It runs in your browser from the site's own files; it is not a company your browser goes off and contacts. That is why this page is about the 84 sites that chose it, not about the library: 1.6% of the 5,294 sites measured, and twelfth of the thirteen technologies broken out here by reach.
Early contact is close to universal on these sites
On 81 of the 84, at least one company was contacted before a consent banner could have been answered — 96.4%, which ranks second of the thirteen technologies in this index. The median count of companies arriving inside that window was 6, also second of thirteen — level with Segment and Hotjar.
Neither number is caused by Alpine. The library makes no third-party request of its own, and the crawl records early contact per site rather than per component, so nothing here attributes any of those seven companies to anything in particular. What the pairing describes is the kind of site that picks a no-build-step library: usually one where the HTML comes from a server, the interactivity is a garnish, and the marketing and advertising code was installed separately and earlier.
First contact lands at 498 ms, the fast end of thirteen
Median time to first third-party contact on these sites was 498 ms. Among the thirteen technologies in this index that ranks thirteenth of thirteen, on a scale where rank 1 is the slowest median — the fast end of the range. The quickest first contact recorded in the group was 85 ms.
Half a second is not much. It is roughly the time between a page starting to arrive and the first paint of anything readable, which puts the first outbound connection level with, or ahead of, the moment a visitor could see there was a question to answer.
Eight companies on the median page
The median Alpine.js site contacted 8 separate companies over 37 requests, with 1 of them classified as tracking against the vendor catalogue. Eight companies ranks fourth of thirteen; 37 requests ranks sixth.
The tracking figure is the one that sits oddly next to the rest, ranking last of the thirteen. A single classified tracker against eight contacted companies means most of what these pages reach has no catalogue entry — advertising hostnames, delivery networks and one-off vendor domains that the classifier has nothing to say about. The count of companies is the more reliable number of the two.
About half the sites showed a banner
Consent software was detected on 43 of the 84. The IAB TCF framework was seen alongside on 20 of them, OneTrust on 12, Cookiebot on 5, and Iubenda, CookieYes and Usercentrics on 2 each. A site can carry more than one of these at once, so these are counts of sightings rather than slices of the 43.
The split matters for reading the 96.4% above. On roughly half of these sites there was never a question in the first place; on the other half there was one, and something was still reached before it could have been answered.
The example sites have nothing else in common
datadoghq.com contacted 24 companies with 22 of them inside the window, no banner detected, first contact at 535 ms. datadome.co: 19 and 18, behind a OneTrust banner. t-mobile.com reached 26 companies but only 6 inside the window, first contact at 951 ms, also behind OneTrust. okta.com: 8 and 8. forter.com: 4 and 4 at 443 ms. spiegel.de contacted a single company, behind an IAB TCF banner. moe.video: 5 and 5.
The rest of the sample is adult video — spankbang.com at 13 companies with 7 inside the window and first contact at 267 ms, spankbang.party at 14 and 12, missav.ws at 11 and 9 with first contact at 196 ms, missav.live at 10 and 8, missav.ai at 9 and 7. A monitoring vendor, a bot-detection vendor, a telco, an identity provider, a German newspaper and a cluster of tube sites. Whatever a stack detection tells you about a site, its category is not it.
| Site | Companies | Before consent | First contact |
|---|---|---|---|
| forter.com | 4 | 4 | 443 ms |
| datadoghq.com | 24 | 22 | 535 ms |
| t-mobile.com | 26 | 6 | 951 ms |
| okta.com | 8 | 8 | 658 ms |
| moe.video | 5 | 5 | 477 ms |
| spiegel.de | 1 | 1 | 583 ms |
| spankbang.com | 13 | 7 | 267 ms |
| datadome.co | 19 | 18 | 297 ms |
| missav.ws | 11 | 9 | 196 ms |
| spankbang.party | 14 | 12 | 375 ms |
| missav.live | 10 | 8 | 498 ms |
| missav.ai | 9 | 7 | 250 ms |
| Tool | Sites |
|---|---|
| IAB TCF consent tool | 20 |
| OneTrust | 12 |
| Cookiebot | 5 |
| Iubenda | 2 |
| CookieYes | 2 |
| Usercentrics | 2 |
Questions
What is Alpine.js used for?
Adding interactive behaviour to server-rendered HTML using attributes in the markup — x-data, x-show, x-on — without a compile step or a component framework. It is small, it is dropped in with a single script tag, and it leaves page rendering where it was.
Does Alpine.js send data to a third party?
The library itself makes no third-party request in this measurement. Every company counted on this page was contacted by something else on the site. Alpine's role here is as a marker for which sites were measured, not as a destination.
Why is the tracker count only 1 when 8 companies are contacted?
Classification depends on a vendor catalogue, and most of the hostnames these pages reach have no entry in it. An uncatalogued domain cannot be classified as tracking or as anything else, so on pages built out of unfamiliar hosts the tracker figure understates by construction.
How many Alpine.js sites had no early contact?
Three of the 84. On those, nothing third-party was reached before the point at which a visitor could have answered a banner.