Sandbox Drama

There’s no more contentious topic in ad tech than Google’s Privacy Sandbox. It has everything! Complex technology. Government oversight. Angry competitors. And a deadline.

This week on the Marketecture pod we went into depth with Alex Cone from Google and talked about the facts, the feelings, and the future. Read more about the interview and Ari’s take below.

Vendor interviews: Aryel and Waterbucket

Editor’s note: Last week we misnamed the CEO of Audigent. Our interview was with Drew Stein.

This week we’re bringing back our award winning* program Justify Your Existence where we interview pre-A round ad tech start-ups and ask them to do just that. These interviews are really short, generally less than 5 minutes.

This week we’ve got two creative vendors that we’re betting are new to you.

Milan-based Aryel is a rich media vendor that is pushing hard into AR and virtual experiences within ads. They’ve enabled immersive ads on a network of sites, apps, and DSP placements.

Waterbucket optimizes the best “buy now pay later” price within retail ads across digital and social channels. They keep track and calculate the best price that will attract customers across millions of SKUs on behalf of large retailers.

* we did not, in fact, win any awards.

Podcast: Alex Cone from Google

Alex Cone, Privacy Sandbox Product Manager from Google came on the pod and we talked for almost an hour about the latest developments, the IAB’s criticism, and more.

Listen to the pod now:

• Spotify

• Apple

• Anywhere you listen

Sandbox Drama

In an earlier podcast I joked that the ad tech community’s feeling about the Privacy Sandbox was following the stages of grief, and had moved past denial (2022) and then from anger (2023), to bargaining (2024). But maybe there’s still a little residual anger left? Let’s ask the CRO of The Trade Desk, Jed Dederick what he thinks:

The IAB Tech Lab, on the other hand, appears to be firmly in bargaining mode. They published a 106 page report (pdf) detailing their concerns. We await Google’s response, which Alex Cone told us on the pod would be forthcoming this week. In the meantime we read the report, so you don’t have to.

The IAB took an expansive view at the Sandbox proposals, covering both purely technical issues and the way it fits more broadly into the marketplace. Here’s a good quote:

In the coverage from AdExchanger, the efficacy of the Sandbox was also put to task:

Let’s break down the basic arguments from the IAB and from others and see what holds water. I’m using “Sandbox” as shorthand for Protected Audiences, and I’m paraphrasing both the arguments from the IAB and the responses from Google’s Alex Cone on this week’s pod.

This doesn’t support all the our use cases. Yes, that’s kind of the point. The criticism of the current status quo is that the uncontrolled usage of 3P cookies has allowed proliferation of all kinds of bad stuff, and it should be expected that some of it won’t be feasible in the new world. One example that comes to mind for me is cross-device graphs.

There’s probably some minimal level of support without which the ecosystem wouldn’t be healthy, and would cause the entire Sandbox effort to fail. And there’s a maximal level of support which might be so complex that it would be impossible to support. The IAB report calls out multi-touch attribution, CPA models, and competitive exclusion as some examples of unsupported use cases. You could probably make arguments either way on whether these should be supported.

There’s also the counter argument that some of these use cases are actually supported by the Sandbox, but in non-obvious ways that still need to be documented. Video support is probably in this bucket. This brings up the next criticism…

The documentation is changing too often, is confusing, and is too technical. Amen, brother. On the other hand, its better than the existing GAM documentation, so there’s that.

The browser is not a counter-party. I’m paraphrasing, but the IAB report makes the point that there’s no one ultimately (financially) responsible when things go wrong. If the browser is the ad server and the exchange, where is the accountability?

This is a pretty interesting argument, but IMHO I wonder how this is different from what we’ve had to navigate through to reconcile discrepancies and cookie match rates, albeit on a much lower complexity scale. Ad tech has always been dependent on browser technology, why is this different?

Audits and accreditation. If you move measurement methodology into the browser how can you be sure it is accurate to the standards necessary for billing? Can the MRC audit a browser’s source code?

When I asked Alex about this issue he essentially said that since its all open source the audit shouldn’t be issue. Is this just a new way the media business needs to think about accreditation? I wonder if there’s been any backlash in the Apple ecosystem to SKAD-based CPA measurement used for billing.

Once again, we’ve had the magical threshold of 10% discrepancies as being OK for 20 years and that hasn’t bothered anyone. Why is this different?

Scalability / You’re running an auction in the browser. This has been the main argument against protected audiences since it was still called TURTLEDOVE. This API can potentially run thousands (!) of auctions in the browser on each page view, with each running arbitrary code.

The IAB brings up the additional question of the impact of resource constraints on ad outcomes, and how that’s auditable and discoverable. For non-techies let me give you a super stupid and unrealistic example — imagine if the auctions ran in alphabetical order so any bidder that started with a “Z” got timed out more often than bidders earlier in the alphabet.

Alex’s retort to this line of thinking is that this is no different really from the situation with pre-bid and RTB timeouts, where the parties can control the constraints they implement. I get this argument, but it also sort of lends itself to a dystopia where the lowest quality sites get even slower and more resource intensive.

There was also a subsequent discussion in X about who gets to set these timeouts, publishers or SSPs, and how there might be conflicts of interest here. I’ll leave this to the reader.

Governance. This is the big one and the space where we start getting the feels. If we, as an industry, widely adopt the Sandbox as the underlying structure of open web advertising we are implicitly giving the keys to the car to our ex who totaled our Kia in college and didn’t even pay for the damage. Or put another way, the future of advertising will be controlled by the world’s largest seller of advertising. A company that is currently being sued by the DOJ for abuses in governance. I’m sure he’ll change!

I asked Alex about two important subjects as it relates to governance.

First, is there a scenario where the Sandbox takes aim at HEMs like UID2? His answer was not entirely satisfying. He said that they work slowly, and that if something like HEMs were in their crosshairs, it would be a documented proposal well before anything would be implemented. OK great, so we’ll know well in advance that we’re totally screwed.

In reality, and as Alex pointed out, Google will be under the scrutiny of the UK’s CMA until at least 2028, so any change to HEMs would be heavily scrutinized.

Second, I pressed him on whether the involvement of the W3C was meaningful in any way considering the indifferent or hostility of the other browser companies. To this he gave me a surprising answer. He sees some movement among the browsers to reconcile or standardize reporting technology. That could mean a future where privacy-protecting marketing reporting works in the same or similar ways across, say, Apple, Android, and web environments. No promises, but a glimpse of hope.

…

Multi-party negotiations in the public spotlight are going to be very complicated. The IAB Tech Lab’s publication has made a great step forward by clearly articulating the concerns of a large constituency, and doing it with thoroughness. Google’s engagement and responsiveness is also really encouraging as we try to get to solutions.

The next two stages are depression (I’m there already), and then acceptance (2025?).

Reading list

Here are some of the big stories from this week:

• The IAB Tech Lab publishers a detailed report criticizing the Privacy Sandbox (AdExchanger article, IAB report)

• Disney, Fox and Warner Bros. Discovery announced a new sports streaming service that could change the media landscape (WSJ)

• Premion acquires DSP Octillion in a potentially big blow to recently-funded MadHive (AdExchanger)

• YouTube now the 4th largest MVPD in the US with 8 million subscribers, passing Dish/Sling (letter from Neal Mohan)

• Expedia runs global campaign on Netflix, illustrating unique advantage of tech over traditional media (Ad Age)

• Google antitrust trial scheduled for September. The judge said a summer trial would have logistical problems, by which we assume he means its just too hot in DC! (Reuters)

• Criteo posts good earnings with revenue flat-ish ($564m to $566m YOY). Retail media now on a $200 million run-rate. (Yahoo Finance)

• Criteo also estimated that the impact of cookie deprecation would be $30-40m revenue in 2H. Back of the envelope, that 10-ish percent? A little hard to calculate exactly but not huge.

• Uber growing like mad in ads. In our Marketecture interview (free with reg) less than a year ago they were at $500 million run rate, and now they are at $900 million. They have 500,000 advertisers - up 75%! - and claim on average adverisers see 8x ROAS.

More Marketecture.tv: