Intent to Experiment: Conversion Measurement API

Skip to first unread message

John Delaney

Sep 17, 2020, 2:31:33 PM9/17/20
to blink-dev, Charles Harrison, Jeffrey Yasskin

Contact emails,,



Spec is very much a work in progress (


This is a new API for measuring conversions (e.g. purchases) and attributing them to clicked ads, without using cross-site persistent identifiers like third party cookies.

Link to “Intent to Prototype” blink-dev discussion

Goals for experimentation

For an initial trial, we hope to see that the measurement data made available through the API produces comparable results to existing click through measurement solutions on the web.

We will be monitoring standard heartbeat metrics and API specific error metrics, such as network issues when sending reports.

Experimental timeline

We plan to start the experiment in M86, and run the trial for 3 milestones, 86->89.

Any risks when the experiment finishes?

This API is only additive and does not affect any existing state. Current conversion measurement solutions will work as expected with and without the API, so we don’t believe there are any risks.

Ongoing technical constraints



There is a chrome://conversion-internals page which allows developers to inspect/modify API storage, and verify that parts of the conversion flow are working as expected.

Will this feature be supported on all five Blink platforms supported by Origin Trials (Windows, Mac, Linux, Chrome OS, and Android)?


Link to entry on the feature dashboard

Third Party Origin Trials

When third-party origin trials are shipped, we intend to add support for them in this experiment. The standard API flow involves two different top-level origins, and in many cases the API is leveraged by a common third party on both of these sites. By allowing third party scripts to enable the trial, this will greatly decrease the burden on developers to enable the origin trial on all necessary sites.

Sep 17, 2020, 3:22:12 PM9/17/20
to blink-dev,,,
LGTM to experiment

This seems like a reasonable way to measure the impact of the new API, and the risk of developers relying on the API in lieu of their current high-entropy measurement mechanisms seems low.

Reply all
Reply to author
0 new messages