Questions about FLEDGE Origin Trial

[appascoe]

Hi, @michaelkleber,

We’ve been in touch with the Google Ads team about one of their recent writeups, by @sbelov, on how AdX may potentially interact with the FLEDGE origin trial. We have a few questions on what the Chrome Team is thinking about; we figured we’d file this issue, but we’d also like to schedule this as a topic for the upcoming FLEDGE WICG meeting.

• What percentage of browsers do you expect to opt in for the origin trial?

We ask because we know that any tests we’re to run on the DSP side will be limited by the participation of publishers in the trial. After discussing internally, and we know this may not be feasible from the Chrome side, we’d figure the best thing would be for 100% of browsers to support interest group bidding so that we can get as much data as possible and perform a cookie split cleanly on our end. This number may not be settled, but any ballpark figure would be useful for us.

• Do you have a sense of which SSPs and publishers are intending to participate in the trial?

This would help us understand on which SSPs we need to be prepared to service FLEDGE-style requests, and also give us a better sense of the scale of the trial.

• When we receive a bid request where FLEDGE-style bidding is supported, will we receive the cookie in the request?

This would be preferable for us, because we could perform a split on the cookie ID space. Essentially, we’re thinking that our population should be “all browsers that support FLEDGE,” and from there, break that population into test and control groups, where we bid through FLEDGE on the former, and classic 3pc bidding on the latter. (We’re aware that these groups may conflict with how SSPs and publishers also decide to split their traffic.)

Or, if this is not the case, would a single auction result in two requests: one flagged as a FLEDGE request and the second business-as-usual?

• It seems like the primary constraints on the test will be 1) browser opt-ins, 2) SSP participation, 3) publisher participation, and 4) DSP participation. Given these constraints, are you targeting any overall volume for the trial?

This is our set of questions for now. We think a broader discussion during the WICG meeting would be most useful so all participants can provide as much valuable testing data as possible, with a shared understanding of the structure of the trial.

Thanks!

[vincent-grosbois]

Hi @appascoe
I have a hard time understanding question 3 and 4: to me, what is intended in the OT of Fledge, is only to expose in the browser the new JS function "adInterestGroup()" and "runAdAuction()".
So the concepts of "Fledge requests" and "Fledge-style bidding" are something that should be discussed with the SSP and publishers participating in the OT, but it's not in the scope of this github repo, which is agnostic on how Fledge is integrated into different other stakeholders.

I agree though that I'd also like to get more infos on how SSP and publishers plan to integrate Fledge for the OT :)

[michaelkleber]

(Initial discussion happened in yesterday's call, notes at https://github.com/WICG/turtledove/blob/main/meetings/2021-09-29-FLEDGE-call-minutes.md)

[vincent-grosbois]

Hi @appascoe
I discussed, I think it's basically not the right space to discuss these topics
Not that it's a problem that we discuss on the repo related to Fledge itself, but to me, the "Fledge integration with adservers" topic is a huge topic that probably requires a dedicated space and dedicated calls etc.
BTW if anyone knows of such already existing "community", I think they should share the link here :)