Clarification on unattributed source/trigger prioritization behavior

[johnivdel]

Hi,

I'd like to clarify (and hopefully align on) some of the specific behavior regarding trigger prioritization and unattributed sources for PCM and Attribution Reporting. One specific scenario I am interested in:

Assume all source/triggers are for a single source site, destination site pair.

t= 0 mins click source -> trigger (priority 1)

t= 5 mins click source -> trigger (priority 2)

In Attribution Reporting, this will generate two reports for a few reasons:

• new unattributed sources are selected for attribution over already attributed sources (see https://wicg.github.io/conversion-measurement-api/#process-an-attribution-source)
• trigger prioritization is only done when there are multiple triggers for the same source (https://wicg.github.io/conversion-measurement-api/#triggering-attribution).

@johnwilander could you clarify:

• will PCM send two reports here?
• does PCM allow there to be multiple outgoing reports for a single (source, dest) pair?

[johnwilander]

Ha, ha! I think we were writing about this same topic in parallel. See if my https://github.com/privacycg/private-click-measurement/issues/88#issuecomment-920310472 clarifies what you're after.

[johnivdel]

I think this is slightly different because there is a redirect trigger in-between the sources.

The behavior of having two outgoing reports at once for a single (source, dest) is also a separate question.

[csharrison]

Friendly ping @johnwilander, can you outline the PCM behavior here?