Download websites that use Google Analytics

About Google Analytics

An analytics tag is not a technology choice in the way a server or a framework is. It is evidence that somebody cared enough about the outcome of a site to measure it. That makes its footprint a reasonable proxy for commercial seriousness, which is a more useful thing than it first appears.

As of September 2026 we detect it on 18,792,464 sites. A detection count of that kind is really a count of properties where someone at some point wanted numbers. Read against the rest of the stack, it separates sites that are being operated from sites that merely exist.

The current generation is GA4, and the migration to it is the part of this data with a short shelf life. The previous generation stopped collecting, so a site still carrying only the old tag is a site whose front end nobody has touched since that deadline passed. Sites carrying both are mid-migration or were never cleaned up. Each of those states says something specific about how actively the property is managed, and the distinction disappears the moment you treat the tag as a single yes or no.

The more valuable signal, though, is not presence at all. It is identity. Because we extract the measurement identifiers rather than only detecting the script, the export supports a question no directory can answer: which properties are administered by the same operator. Domains sharing a measurement account are almost always run by the same team, and that mapping cuts across brands, registrars and hosting arrangements that otherwise look unrelated.

That single capability changes the shape of a lot of research. Competitive analysis stops at the brand boundary without it. Partner and reseller mapping is guesswork without it. Due diligence on a portfolio of web properties is substantially harder without it, because ownership on paper and ownership in practice frequently differ.

Geographically United States accounts for 8,631,408 of detections and the .com zone leads, which follows the commercial adoption curve rather than any regional preference for the tool.

The export carries the identifiers alongside the domain, the platform, the hosting country and the rest of the detected stack, as one file rebuilt daily. Grouping by identifier is a local operation over the whole population, which is the only way that particular query is affordable at this scale.

A caution about the identifier data is in order, because it is powerful enough to be misused. Shared identifiers are strong evidence of shared administration, not of shared ownership. Agencies, hosting companies and marketing partners routinely deploy their own accounts across client estates, so a cluster can represent a supplier relationship rather than a corporate one. Treated as a lead, that is still valuable. Treated as proof of ownership, it will occasionally be wrong.

The absence of any measurement tag is also a signal, and an underused one. A commercially active site with no analytics of any kind is either very new, very neglected, or deliberately privacy conscious, and each of those is a small and identifiable population. You can only ask that question if you hold the sites where nothing was found, which is a row that most per-technology directories simply do not return.

Google operates it as a Proprietary service and it has been available since 2005. Because it is free to deploy and requires no relationship with the vendor to appear on a page, its presence tells you about the site owner rather than about any commercial arrangement.