Download websites that use Rubicon Project

About Rubicon Project

Names in advertising technology outlive the companies that carried them. Rubicon Project merged into Magnite and the corporate identity changed, but the tags, wrappers and bidder identifiers deployed under the old name kept firing on publisher pages, because nobody rebuilds a monetisation stack for a rebrand. So a detection under this name is not only a statement about who sells the inventory. It is also a statement about how long ago the integration was set up and how rarely it has been touched since.

That makes the signal useful in two directions at once. Read forwards, it identifies publishers running programmatic demand through a supply-side platform, which means a site with inventory, an ad operations function of some kind, and a revenue relationship worth competing for. Read backwards, it identifies ad stacks that have been left alone, which is the population that header bidding vendors, yield consultants and rival exchanges spend real money trying to find.

As of October 2026 we detect it on 228 sites in the Advertising category. A figure in that range should be read carefully, because the supply side of advertising is intermediated. A publisher does not usually decide to add an individual exchange. A managed service, a network or a wrapper configuration adds it on the publisher's behalf, so detected domains tend to arrive in clusters that share an intermediary rather than as independent adopters. For prospecting, the cluster is often the real unit and the individual domain is the noise.

Magnite Inc. operates the platform, which has been in the market under a Proprietary model since 2007 and grew up alongside real-time bidding itself. Longevity matters for interpretation. An exchange integration that predates the current generation of tooling frequently coexists with newer tags on the same page, and the layering is visible in the data: sites carrying several demand paths at once, sites where one path has clearly replaced another without the old one being removed, sites where the remaining tag no longer earns anything for anybody.

Detection has firm limits here and they are worth stating plainly. We see what is present in the page a crawler receives. Tags injected conditionally, delivered through a tag manager after consent, restricted to certain geographies, or bundled inside a wrapper that does not name its bidders in the markup will be under-reported. Our figures describe visible, unconditional deployment, which is a lower bound rather than a measure of real spend.

The value of the export is that it puts every one of those domains next to everything else fingerprinted on the same site. Advertising stacks are only interesting in combination, because one tag tells you almost nothing while the set of tags on a page tells you who the publisher works with, what generation of tooling they are on, and where a competitor could realistically displace someone.