W3C

– DRAFT –
Improving Web Advertising BG

02 November 2021

Attendees

Present
AramZS, blassey, brendan_eyeo, charlieharrison, dmarti, FredBastello1, johnwilander, jrosewell, Karen, kleber, kris_chapman, lionel_basdevant, mjv, nics, tonytan, wbaker, weiler, wseltzer
Regrets
-
Chair
Wendy Seltzer
Scribe
Karen, Karen Myers

Meeting minutes

<wseltzer> experimentation

<wseltzer> Lab's Content Taxonomy 3.0:

<wseltzer> https://iabtechlab.com/press-releases/tech-lab-releases-content-taxonomy-3-0/

<wseltzer> what should be in-scope for charter(s)?

<wseltzer> Click tracking proposals & TURTLEDOVE / PARAKEET proposals

<wseltzer> proposal status

<wseltzer> experimentation

<wseltzer> Lab's Content Taxonomy 3.0:

<wseltzer> https://iabtechlab.com/press-releases/tech-lab-releases-content-taxonomy-3-0/

<wseltzer> what should be in-scope for charter(s)?

<wseltzer> Click tracking proposals & TURTLEDOVE / PARAKEET proposals

<wseltzer> proposal status

<nics> present

Introductions and Agenda Curation

Wendy: These are agenda items carried over from the TPAC meeting requests
… usual intros, reporting on experiments

[Wendy reviews agenda]
… Among the agenda items, IAB Tech Lab is available in a few weeks to present on contextual advertising...move to Nov. 30th
… for that conversation
… thanks in advance to them for information sharing there
… Among other subjects
… Do we have particular interest or availability to share updates?
… at the TPAC meeting we had a vigorous discussion of the privacy signals
… and in the other meeting we discussed metrics and some possibilities
… overview
… we still need to do more work to figure out cross time zone meetings more effectively
… a lot of groups are looking at that

<seanbedford> +present

<tonytan> +present

<jrosewell> Still reading minutes re: signals - did this cover all the x4 options?

Wendy: Do we have any introductions? Anyone new to the group who would like to introduce themselves?

Tony: I am at Apple
… I recently joined the company; happy to join the group and nice to meet you all

Jesse: I am new to the group, I'm over at Xaxis
… work on ops and tech side teams

Alexandre: I am with IAB Europe

Wendy: Welcome
… Did we have any reports from experiments or origin trials that people wanted to share?

Reporting on experiments: learning from origin trials and other

Wendy: We had a good discussion a few weeks about the Machine Learning experiment that Criteo had run
… Anything that we want to bring into this group from experiments, the platform, or thinking about other experiments that might be underway

Charlie: I can give a quick update on the origin trial that Google Chrome ran
… and has restarted for the attribution API
… we ran origin trial for click-through conversions...
… get a report that tells about join behavior across these two sites
… We got a bunch of great feedback
… I'll link summary in minutes
… John Delaney put together
… We got formal feedback from Yahoo Japan
… we learned interesting stuff on new paradigm of doing attribution on device

<wseltzer> https://github.com/WICG/conversion-measurement-api

Charlie: One big thing we learned is that when you do on-device attribute
… there is more control of data being stored in browser
… We saw users deleting data for the API, maybe sensitive
… clear that data, or clear the browser history
… that destroys the attribution data which otherwise would be preserved of third party cookies
… consequence of doing things on device
… We did not expect so much impact

<charlieharrison> https://groups.google.com/a/chromium.org/g/blink-dev/c/DdjaFmsb4fA/m/RTK45f9gBQAJ?utm_medium=email&utm_source=footer

Charlie: We also learned interesting things with network reports, retries, technical details we need to nail down for these APIs to work
… I will post in irc
… a bunch of links to other feedback, and issues that we have filed in doing the origina trial

Wendy: thanks, Charlie

Aram: this is really interesting results
… users are doing manual clearing behavior?

Charlie: That's right
… it's not that easy to clear third-party cookies
… many are set a while ago, months and months ago
… common thing people do is to clear recent browsing history
… might expect this to be the same but it's not
… that cookie is still around
… also site data clearing
… for example, won't clear all cookies
… third party cookie deletion might be expected but it's not
… new window of time where deletion is possible
… Three events: impression, logs to storage; conversion, logs to report
… as soon as conversion happens, you have attribution
… but because of this delay, user can delete the conversion report

<charlieharrison> https://developer.chrome.com/docs/privacy-sandbox/attribution-reporting-data-clearing/

Charlie: We have a blog post that goes through these nitty gritty details
… I think that is linked from link in irc
… I'm copying it here
… Goes through these cases and shows how this is different from third-party cookies

Aram: curious if there has been looking at this activity here
… that users are expecting stuff to be cleared
… and reflection that this is data....not being cleared with cookies

Charlie: could clarify user intentions
… with Chrome, could image us changing clear site data to have those semantics if that was user intent
… We don't know enough about it yet to know if that is the case
… I definitely think it's an interesting result
… in terms of privacy sand box
… speaks to doing more stuff on browser and gives users more control basically

Aram: very interesting

John Wilander: Wanted to mention
… another way to look at users clearing data...is partitioning
… you will be clearing for first party web site
… may meet user expectations more
… I don't think it's possible to meet user expectations without partitioning
… may log users out of some services
… to explain
… because those cookies are double-purpose, used for log-in or whatever

Wendy: If people want to follow up conversations on that there is work on Privacy CG on partition storage
… and work in Federated ID community group thinking about the sign-in situation should look like, with or without third party cookies
… and we heard on Friday last week, the Private Adv Tech group has just launched and is taking proposals to be incubated as well
… the queue is empty, thank you for sharing all those links, Charlie
… A lot to read up on the origin trials

<Lisa_Markou> No

Wendy: as people are interested to review and read more

<Lisa_Markou> Sorry nothing to say actually

Preparing for the standards track: what's ready to move to a WG,

<blassey> for future reference "q-" will take you off the queue

Wendy: preparing for the standards track and thinking about what might be ready to move to a working group
… this was a question from reflection among the work that's being incubated
… and ask whether there are places that people see working groups in the likely near future
… as we saw the PAT CG

<wseltzer> https://www.w3.org/community/patcg/

Wendy: which just launched with a charter specifically
… including the goal of drafting a charter for a working group

s/group
… or circulation to community for proposal to move work forward
… that is one place where we expect to see the community identify where there are proposals not only of interest to multiple segments of the community
… but to have convergence on a proposal that would be put forward a specific work item to charter
… I don't know if there is more people want to say here
… in that discussion
… in that path towards chartering
… or if we are waiting to see how incubations fare
… a goal in PAT CG is to be a more focused discussion
… including taking things that have been incubating in WICG

<charlieharrison> sorry folks, I have to leave early today. Thanks!

AramZS: I am one of the chairs of the PAT CG group
… want to say hello since it came up
… link to group is in irc
… anyone can join if you are a W3C member
… we are already planning meetings

<wseltzer> https://patcg.github.io/

<wseltzer> https://github.com/patcg

AramZS: anyone interested in participating in bringing things toward standards should join in
… Chime in through Github, linked from that page
… Any questions, just let me know
… I'm available on Twitter, irc, and on PAT CG mailing list; just let me know

Wendy: thank you, Aram

<jrosewell> Aram; when is next meeting?

<AramZS> Next meeting is not yet set

<AramZS> We're setting up the next meeting time in the Github and when the timing is set we'll also send out an email to all participants.

Wendy: update on reconciliation processes for proposals that had multiple tracks

<Lisa_Markou> Is someone from IAB here that may be able to share more on id-sources.json

Wendy: aggregate measurement, @, PARAKEET and Turtledove proposals
… proposals aiming at similar use cases but not identical mechanisms
… trying to do the same thing
… Do we have possible convergence towards web-wide standardization
… Don't know if participants in those efforts have updates to share or if there is work to be done?

Angelina: I am not involved in that WG
… but since term 'reconciliation' did come up
… when it comes to reconciliation, buyers and publishers
… typically require currently that all data be reported as of end of the month
… so that billing for prior month invoice is accurate
… that is what they agreed to
… With delayed reporting, will there be an opportunity
… most of work on CPM
… or deliver basis
… at end of month so billing doesn't get screwed up
… if so, adtech goes back to buyer and say you owe more money due to more impressions

Wendy: Thank you
… interesting to note two different uses of the term reconciliation
… might have referred to bringing different proposals, or to refer to a consolidated measurement at the end of the month
… Charlie apologized that he had to leave

MichaelKleber: thank you for pointing out the different meanings with the term 'reconciliation'
… I can respond in part to Angelina's question
… the various aggregation proposals do have some amount of turn-around time in how long it takes to perform aggreation
… how long it takes to move through system, more than one day of lag
… end of month reporting need would not be a problem regarding any of the aggregation APIs we have talked about
… what Charlie mentioned on conversion reporting

<anderagakura> @Liza_markou : regarding id-sources.json = - Id-sources.json was developed to enable companies to declare the user identity sources they use. - Participants who are integrated with identifiers from different providers can quickly recognize the supply paths in which they can activate addressable audiences

MichaelKleber: there is a longer delay
… conversion reports are sent on the end of a few different pre-set time windows
… some amount of time after the original click event happened
… something like 2, 7, 30 days as default, although I might be mistaken
… if you are looking at an end of the month number based on how many conversions have happend
… might be some substantial delay if event happened 7 days post click, it might take to end of 30 days before it's reset
… that is what Charlie mentioned in time when users delete data
… also when you know number of conversions in a particular month

Angelina: 7 days is tolerable for most buys
… that latency window would have to kick in
… most people rely on end of month
… whenever attribution happens
… if user decides not to open the browser in the 2-7 bus days after, that level of data would get lost?
… or would it get re-aggregated the next time that browser opens
… if browser is not active, will it still send the signals

Michael: Right
… Charlie and @ are best to answer those questions
… high level answer is the browser would try...reports sent when browser next opened, but a risk info might be recorded later

<wseltzer> https://developer.chrome.com/docs/privacy-sandbox/attribution-reporting-data-clearing/

Michael: that blog post does have some info about this question

Wendy: Great; that exchange makes me think about the use cases
… to make sure we have captured in their accounts of the various reporting use cases
… and the time periods that are useful for information capture in any particular needs there
… so we have a reference point for looking at
… how we are able to satisfy various reporting requirements
… I want to encourage again
… people to look at that use case document
… and think about places

<wseltzer> https://github.com/w3c/web-advertising/blob/main/support_for_advertising_use_cases.md

AOB

Wendy: now that we've had lots more discussion
… do we have issues to raise, pull requests to add, new material there
… if anyone wants to engage
… in some documentation sprint activities
… queueing up sections of that document
… for another read
… and more thinking about how we've captured the use cases
… and where we now see proposals meeting those use cases
… and continue following those in the detailed discussion of the proposals
… That's what I had in mind with the documentation sprint agendum
… And that takes us down to...

Lisa: I do have a question
… do we anticipate with that conversion window changed
… potentially the source of truth for conversion reporting and/or billing becoming a first party pixel
… coming from first-party with no delay?
… Want to get some feedback from the crowd on that?

JohnW: thank you

<johnwilander> https://github.com/privacycg/private-click-measurement/issues/95 (Supporting multiple conversions for one click, a.k.a. reporting windows)

JohnW: I just wanted to highlight
… posting link
… I recently posted on private link measurement
… the click through version

<AramZS> There is a standalone conversions meeting, right? Might be good to link to the current group if so

JohnW: we are looking into multiple reporting windows similar to Google
… also contains privacy analysis with multiple windows one initial click
… and going into reporting delays
… I welcome thoughts on these proposals
… thanks

<AramZS> Stand alone conversion measurement meetings are here - https://github.com/WICG/conversion-measurement-api/issues/80

Wendy: Thank you for that link
… Aram points out standalone meeting on attribution and measurement reporting
… adds a link, thank you, Aram for detailed followup
… that Github repository and the meetings they are scheduling is a good place to follow up and to find the people most likely to have most direct response and engagement on those questions
… thanks for the pointer
… any other business?
… I think it's terrific that we are seeing so many different groups and side meetings picking up on individual topics
… that is serving well
… we do the high-level discussion and needs identification here
… and then there are in-depth exploration of those proposals and detailed work on drafts and follow-up in these additional meetings
… if what we are doing is pointers-out to those other meetings
… and helping to share the follow-up, then I hope that is useful to people in finding
… the current state of work
… and following the pieces that are of most interest
… If we don't have any other business today
… take a moment more for the queue

Wendell: thanks, Wendy
… there has been a bunch of discussion, quips on frequency and intensity of meetings
… we once had this group Adv meeting once every other week
… wondering if that is a better cadence now in the 'interregnum' period
… with overhang of markets
… getting into holiday season
… just throwing out there it might be better to have a slower cadence [for the Adv BG]

Wendy: Thank you
… taking cue from activity in other places
… I am happy to move this to an every other week schedule and adjust the calendar invite to put onto people's schedules
… if this pick up
… we can rearrange further
… happy to do that

<jrosewell> far too many meetings with overlapping scopes and agendas - hard for smaller participants to keep up. Would be good to resolve the core questions like "what do we mean by privacy?" or "how do we measure an improvement?" in one place.

Wendy: that matches my sense of timing as well
… that would put our next meeting here on the 16th and 30th of November
… any other comments or questions?

Angelina: I did want to inform everyone of the task force we created at IAB
… the browser ads testing Browser OS Ads Testing Task Force
… intended for business side
… for people not as informed on all the proposals
… we have had two meetings; had 80 people
… we went over privacy, FLEDGE, FLoC, reporting

<Lisa_Markou> Thank you Angelina! That would be super helpful. Would love it.

Angelina: I can make that report available; just email me
… one thing that came out of TPAC
… that Josh Koran talked about from Criteo is doing KPIs
… date TBD
… we will bring the buyside perspective on what kind of KPIs they are looking for

Wendy: thanks, Angelina
… I think that wraps things up for today
… please continue to share requests for discussion
… I'll see what comes up for November 16th
… See you then

[adjourned]

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).

Diagnostics

Succeeded: s/gorup/group/

Succeeded: s/3 days/30 days/

Succeeded: s/task force/Browser OS Ads Testing Task Force/

Succeeded: s/Josh/Josh Koran/

No scribenick or scribe found. Guessed: Karen

Maybe present: Alexandre, Angelina, Aram, Charlie, Jesse, JohnW, Lisa, Michael, MichaelKleber, Tony, Wendell, Wendy