Download websites that use Rudderstack

About Rudderstack

A customer data pipeline is bought by a data team and not by a marketing team, and that distinction explains most of what the numbers here can and cannot tell you. Rudderstack collects events in the browser and routes them to a warehouse first, then onward to whatever analytics and marketing tools the company runs, inverting the usual arrangement where every tool collects its own copy of everything.

The project appeared in 2019, runs on Go on the server side, and is published by RudderStack Inc. under the Apache 2.0 licence. Open licensing is why self hosting is common, and self hosting is why our count behaves the way it does. A company running its own data plane serves the collection endpoint from its own hostname, and once that endpoint is first party the traffic looks like any other request to the site's own domain. As of October 2026 we detect Rudderstack on 36 sites, and the figure leans towards deployments that kept the recognisable defaults.

Undercounting matters more here than it would for an ordinary Customer data platform tag, because the deployments that hide best belong to the larger engineering organisations, which are also the ones a researcher most wants to see. We publish the detections we can support and state plainly that the population beneath them is bigger than the count.

What a detection does tell you is unusually specific. A site loading a warehouse first pipeline has a data warehouse, has someone who owns event schemas, and has already decided against letting each vendor tag collect independently. Very few single scripts imply that much organisational maturity behind them.

Licensing has a commercial consequence that anyone doing market sizing should hold on to. Under Apache 2.0 terms the same product can be self hosted for nothing or bought as a managed service, and a detection does not distinguish the two, so the count is a measure of adoption rather than of revenue or of customers. A vendor reading it as a customer list would be reading it wrongly, while a vendor reading it as the pool from which paying accounts are converted would be reading it correctly.

Competitive intelligence is the obvious use, since the direct alternatives are detectable in the same dataset and can be lined up beside each other. Warehouse vendors, reverse ETL products, consultancies and the analytics tools that sit downstream all want the same list, because a company that has built this pipeline is in the market for everything that plugs into it. A per domain check confirms one prospect. An export gives you the segment, filterable by country, top level domain and the rest of the stack, which is what account level targeting actually runs on.