Day 3: The problem with header bidding is it exists

The Monopoly Report is going daily during the trial. Our intrepid founder is in a windowless room without a phone or laptop to get you the good stuff.

Day 3: The problem with header bidding is it exists

The title of this email and some other juicy quotes come from Brad Bender’s testimony, where the DOJ used the opportunity to bring to light some emails that probably shouldn’t have been sent. We got our first expert witness and a guest appearance by The Trade Desk.

Here’s what happened.

Chalice is the founding sponsor of The Monopoly Report

Stuff we learned

  • TTD forced investors to watch a video about header bidding.

  • Literally no one can explain what “first look” is and how it relates to Dynamic Allocation.

  • If you try to get a Google Doc into evidence the comments will be seen as hearsay, or the dreaded “double hearsay”.

The podcast tracker — we’re down to 50%!

Plaintiff’s witness 8: Brad Bender, former Google VP

Editorial note: I consider Brad a close friend, but will try to be objective in the name of journalism.

The DOJ focused on three things in bringing Brad to the court:

  1. Discussion of a pretty damning strategic document based on a talk by former DoubleClick CEO, David Rosenblatt.

  2. Discussion of a pretty damning email about header bidding.

  3. Showing the court that Googlers are dirty, lying, rats that can’t be trusted to preserve documents.

David’s strategic talk

David Rosenblatt was the CEO of DoubleClick through the Google acquisition and stuck around in a senior, but non-operational role for maybe a year afterwards. He had a key role in articulating the overall display strategy and helping integrate the company. David is an excellent extemporaneous speaker and during this time he gave a number of talks to customers, employees, and others describing how Google was going to absolutely dominate … er … kill … er … compete … in display.

One day, a nice young man named Clay Bavor, who apparently can type quite quickly, decided to transcribe one of these talks, and the rest is evidence. Here’s a copy of the filing, that is not precisely the one presented in court today, but likely with the same text.

Here’s an excerpt saying you can control inventory in the ad server and there are huge switching costs:

Here’s an interesting comments about the exchange. Basically if the network (GDN) isn’t successful then Google still wins with AdX:

And, of course, we need to wrap things up by “crush[ing] the other networks”.

The header bidding email

The second document that Brad was interrogated about (despite being the author of neither doc), was this frank discussion of header bidding between Brad, publisher leader Jonathan Bellack, and Payam Shodjai, who, as a matter of trivia, used to work for me.

Like a good PRD, let’s start with the problem statement. Bellack says DBM is bidding into header bidding, which is basically the equivalent of wearing a Chappel Roan shirt at a Taylor Switch concert.

Payam, who was the lead PM on DV360 at the time, then let’s it all hang out. He says, in short order, a) header shouldn’t exist; b) header only exists because of Dynamic Allocation; c) we should discuss DBM (DV360) stopping all buying on third-party exchanges; d) Exchange bidding is cool, but the 5% take rate isn’t; and e) Poirot is boosting spend on AdX, presumably at the expense of other exchanges. I am trying to imagine the DOJ employees dancing on their little cubicles when this email came through in discovery.

Google employees suck, especially Brad

The DOJ attorneys really laid into Brad, regarding his truthfulness in testimony and deposition, and especially in regards to alleged improper use of privilege and potentially improper destruction of chat evidence. Here’s an article covering the history of this allegation.

While I have to say that Brad wasn’t exactly forthcoming in his testimony, claiming, for example, that he doesn’t know how header bidding works, I also didn’t see any real evidence pointing to obstruction of evidence, etc.

Google losing share

On cross, Google brought up a chart showing Google losing market share in the “display” market, primarily to Facebook. The chart was in a deck meant for executives, and showed Google’s share declining from the maybe 24% to the low 20’s, while Facebook increased to 56% by 2020. There was no clarification on the sources, or whether this was US or global. Expect to see a lot more of this.

Plaintiff’s witness 9: Professor R Ravi, Carnegie Mellon University

Our first expert witness, Professor Ravi, was tasked with analyzing the various alleged auction manipulations by Google, including “First Look”, “Last Look”, Poirot, and Unified Pricing Rules.

I have first and last looks in "quote marks” because these are not actual products, but rather the end effect of Dynamic Allocation. It seems like neither professor Ravi, not either side’s lawyers, nor several other witnesses understand this.

Some points the professor landed:

  • In general, all of these auction manipulations hurt publishers.

  • Because of “First Look” Adx won 53% of DFP auctions.

  • There is a feedback look whereby the more auctions Google wins, the lower the waterfall prices become for competitors, which makes them win less.

  • “First Look” became “Last Look” when header bidding arrived, but it had the same effect.

  • “Last Look” costs publishers money because it takes auction pressure off of AdX. In other words, if AdX didn’t know the price it needed to win, it would bid higher.

Effect of Unified Pricing Rules

During the professor’s testimony a document was shown estimating the effect of Unified Pricing Rules (“UPR”). Under UPR, Google forced DFP customers, against their will, to put AdX on the same playing field as other exchanges. The doc showed that 42% of auctions won by header bidding had an AdX-specific floor price that was higher than the winning bid. This means AdX couldn’t compete. And of those 42%, 40% were in auctions where AdX had a max bid higher than the winning bid. So basically UPR should increase AdX win rates by ~16%.

There was also an email cited by Google saying UPR only increased revenue for Google by 1-3%, until, in a Perry Mason moment, the DOJ lawyer suggested we look at the immediately following paragraph that revealed the actual effect to be 40%. Doh.

Cross examination scores some points

Google’s cross of the Professor was semantic and terrible, but they made a couple of good points. While Google’s defense is largely focused on the market definition, they also make the argument that the changes they made were for legitimate business reasons. The cross made some interesting points:

  • UPR reduced errors, gives bidders better signals.

  • Dynamic Revenue Share helps both publishers and advertisers (by clearing more).

  • Poirot (bid shading) was tested on AdX and didn’t see much difference in price, presumably because AdX was running a fair auction.

  • Wouldn’t Google enjoy scale advantages just by building good products?

  • Didn’t project Bell (punishing first look deals outside of AdX) actually reduce multi-calling, which is a wasteful practice?

Plaintiff’s witness 10: Jed Dederick, CRO, TTD

I was wondering about the strategy of bringing a witness from The Trade Desk, given that DV360 is not part of the complaint or alleged to be a monopoly. Jed’s testimony seemed to be aimed at showing that AdX’s activities were not in the interest of their largest counterparties.

For a company that’s an industry leader with a $50 billion+ market cap, Jed painted a sad picture:

  • DV360 has numerous advantages over TTD including data, YouTube, and resourcing.

  • TTD hasn’t been able to meaningfully grow in the open web display market despite investment.

  • Every TTD customer also has a Google contract.

One of Google’s defenses to the way it treats external exchanges is around quality. Jed took a blowtorch to this argument , saying “Google and Adx are the worst purveyors of MFA sites in the ecosystem.”

On Google’s continued quest to say that social is a form of display: “I don’t know of any advertiser that would think of these as interchangeable”.

And then, kind of out of nowhere, Judge Brinkema said “[regarding these proceedings] If scale is critical and Google is ‘blown apart’ would that make it more difficult for buyers?” I don’t know, this is a good question.

Finally, and I found this amusing, Jed said that during TTD’s IPO roadshow they made every investor watch an explainer video about header bidding, since they thought it was that important to their future.

Editorial note: I left before cross was finished, so might have missed a bit.

Reply

or to participate.