Day 2: Stephanie is being emotional and unproductive

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 2: Stephanie is being emotional and unproductive

There’s nothing more insulting you can say to a smart, accomplished, professional woman than that she’s being emotional. But apparently that’s exactly what the (female) account team at Google told well-known ad tech guru and News Corp exec Stephanie Layser when she wouldn’t stop complaining about the launch of Unified Pricing Rules. “Emotional and unproductive” for the record.

This, and a bunch of other amazing out-of-context quotes await you in today’s summary of US et al v Google.

Stuff we learned

• During the “last look” era you could trick AdX into bidding higher by simply increasing the bid sent from prebid into DFP.

• The Goodway group’s Jay Friedman wrote two books, and a bunch of Quora answers.

• The politics at Google are absolutely brutal (we knew that already).

Also, its all about me

Where are we at?

Between yesterday’s testimony by DFP competitor James Avery and today’s by publisher expert Stephanie Layser, the DOJ has really beaten home the point that DFP is a sticky monopoly that is highly tied to Google AdX demand, which is largely based on Google Ads and GDN demand. I don’t think any of the cross-exams by Google have put much of a dent in these points.

The stories that the DOJ is now starting to broach are about header bidding, and the ways Google attempted to maintain their exchange power and high take rate. These are not nearly as fleshed out yet.

Google continues to put most of its efforts into market definition. If they can convince the judge that “rectangles on webpages” include “rectangles on facebook.com” then they’ve won.

Plaintiff’s witness 5: Stephanie Layser from Amazon, formerly News Corp

Stephanie is a verified expert in all things publishers. Per her testimony, she’s set up 75-80 publishers with header bidding stacks, has served on the board of prebid and is currently a subject matter expert at AWS.

Project Cinderella

News Corp did a comprehensive evaluation called “Project Cinderella” of whether it was possible to switch off of DFP and onto AppNexus (now Xandr). The findings:

• Revenue risk from losing AdX demand was too high to justify the switch.

• AdX Tags, which are the option for getting its demand on other ad servers, do not bid with a price. Since they just buy or don’t buy, this makes it impossible to have demand compete with other sources.

• They considered a two ad server model, but it would have broken forecasting, cost twice as much, etc.

• Not being able to see AdX bids also hurt innovation since News couldn’t implement a “smart paywall” based on CPMs.

We also learned that in 2016 AdX was 52.7% of News Corp’s programmatic revenue, and 41.9% of that was from Google Ads.

Give her the logs you cowards

Now you might have gotten upset about the “emotional” tidbit above, but the part of Stephanie’s testimony that got my blood boiling was the part about logs. She testified that Google refused to give News Corp matchable logs to compare DFP and AdX after years of requests, and when they finally gave in, they had removed the match key. Give this woman her logs! 🪵 

Direct vs programmatic

Stephanie testified about how she sees the difference between direct and programmatic demand:

A world without AdX demand

In 2019 News Corp did an experiment to suppress AdX demand in DFP when there was a bid coming in from a header wrapper. While AdX revenue declined 37%, the other exchanges made up this revenue, and the overall effect was neutral. This line in inquiry was cut short by Unified Pricing Rules since DFP no longer had fine-grained control over demand, and there was no way to reduce Google Ads demand without also penalizing DSPs bidding through AdX.

Other parts of here testimony:

• Header bidding always increased pub revenue 25-50%. In the early days there was some latency increase, but no increase in bad ads.

• “Last Look”: She ran the analysis and Google consistently would win only 1c more than demand from header, so didn’t make publishers any substantial amount of money.

• “UPR”: “I felt like they were holding us hostage”

• “Minimum bid to win”: This is referring to loss notifications, where after first price auctions were established Google would give AdX bidders info about the clearing price, but would not give that data to header bidders.

The cross examination covered many of the topics above. The more interesting parts of the cross came when some old op eds were dug up. One had a phrase where it appeared that Stephanie was recommending that AdX be donated entirely to Prebid and become open source, which prompted the judge to ask if she believed that ad exchanges should be run by capitalists. (Note, this would not be the last time socialst ad tech would be brought up today).

I’ll leave you with another banger of a quote in answer to a question about what she was talking about when she said she fears manipulation of the auction by Google:

Plaintiff’s witness 6: Jay Friedman, CEO Goodway Group

If Stephanie’s testimony was like a true crime podcast, Jay’s was like listening to NPR at 0.5 speed. Jay was a good witness and is an expert, but it was just a little…dull.

Jay’s testimony was mostly focused on shoring up the idea that media planners and buyers think of different channels differently. He was asked about whether PG was a substitute for open bidding (no). Whether ad networks were a substitute for programmatic (no). Whether facebook is a substitute for open web (no).

Jay spoke about Goodway’s extensive SPO efforts and indicated that he had obtained fee reductions from every exchange except AdX, which was unwilling to negotiate at all. He further testified that he cannot shut off AdX because of concerns about scale.

The cross-examination was, to be totally honest, horribly dull. The Google lawyer surprised everyone (especially Jay) by revealing that he had a book, “30 Days to Paid Digital Media Expertise”. Buy it on Amazon.

There were various interesting quotes in the book, and on the Goodway blog, and, most surprisingly on Quora, that helped Google’s argument about market definition and size. For example, this zen koan took an hour to parse:

Points the cross made based on Jay’s previous writings:

• Google DSP is very competitively priced.

• Some advertisers may use both DV360 and Google Ads.

• There’s nothing to stop advertisers from using Google Ads and other ad networks.

• “Variable Pricing Floors” were “unfortunate” for buyers.

• Header bidding had some latency in the early days.

One last thing because I think its a great illustration I’ve never heard before. To rebut a Google point that all display ads are top funnel, Jay made the point that if a well-known brand like Heinz just puts their bottle in a banner ad it can raise awareness, but if a vodka brand you had never heard of did the same thing it would be a waste of media dollars. Wisdom.

Plaintiff’s witness 7: Eisar Lipkovitz, former Director of engineering, Google

Eisar’s testimony was delivered by transcript and video, and it had some real zingers. He led the overall engineering team of 2500 people from 2014-2018 and seems to have endured some of the more difficult political moments in the push-and-pull over what to do about header bidding.

Like many Google executives, he said some really smart things and some really clueless things (IMHO). Asked about “last look” he called it “a publisher friendly feature since it makes them more money. But it is a very complex subject and hard to say if it is good or bad.”

AWBid

In 2014 Google developed AWBid, the ability for Google Ads to bid on other exchanges. This ability was limited by policy to only remarketing campaigns, which were a very small portion of total GDN/Ads volume. Eisar testified that there were inventory quality concerns, but also financial concerns about losing AdX margin. Eisar quoted Jonathan Bellack as saying “[AdX demand] is a winning argument for DFP”. Eisar testified that bidding on other exchanges would increase bid densiity and competitiveness of other exchanges.

Without naming names, Eisar painted the picture of painful, drawn out political battles to resolve the future of AWBid. “I have PTSD” was one such comment.

Header bidding and Open Bidding

Lots of thoughts on these topics:

• Eisar explained various problems with header bidding, such as latency, billing issues, and advertisers bidding against themselves.

• He characterized GDN as having unique demand because “we don’t care if it is USAtoday or some other site, as long as we have the signals we need.”

• Last look was only removed after negotiations with SSPs to get them to adopt Open Bidding.

• AdX Direct: this was political and got no traction.

• Poirot (reducing bids on non-AdX exchanges) wasn’t meant to help AdX, it was meant to help advertisers.

This brings us to our second socialist ad tech thought of the day:

AdX take rates and “drain the swamp”

The AdX/GAM team in a series of emails and memos outlined ways to fix the mess everyone made with header bidding, 2nd price auctions, open bidding, etc. While we haven’t seen the whole emails yet, one topic of discussion was reducing the AdX take rate significantly. This was a political hot potato and Eisar expressed a lot of frustration that it wasn’t really resolved.

I’ve got to admit to being a bit confused about this topic, because if AdX was cheaper (after 1p auctions and removal of last look), they would just win even more, and that wouldn’t stop anyone from doing header bidding.

What’s next?

That’s all I’ve got. The next couple of days will probably turn to header bidding, Poirot, and some other topics.