Day 14: My market is bigger than yours

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 14: My market is bigger than yours

Today was almost entirely dedicated to Google’s final in-person witness, Dr. Israel, who presented what I’d describe as avant-garde interpretations of the ad tech market, including that Facebook was really good at SPO and that if you don’t like DFP you should get your customers to download an app instead. There’s also a cameo from the Costco chicken.

Here’s what happened. Sorry, this is a long one!

Stuff we learned

• Google’s display business wasn’t profitable until 2018.

• 21% of DV360 spend goes to YouTube.

• The Google display business spends $1.5 billion on engineers every year.

Google witness: Jessica Mok, Google

Jessica is a finance director at Google, and from 2017-2023 she focused on the display business. Google brought her to the stand to show that the business was not exactly the cash cow you expect, which they are hoping will hurt the monopolist argument.

For whatever reason, they disclosed the data in the chart below from 2015 through 2019, indicating that the display division (DFP, AdX, AdMob, AdSense, GCM, GDN) was burning cash, then became much more profitable in recent years. These numbers do not include any buys on YouTube or Search. The 2020 and ‘21 numbers are extrapolated from an exhibit where they claimed to be “on a $3 billion run rate in 2021, which is 166% higher than last year”.

Jessica also gave us some color on these numbers:

• 50% of the profits were from AdMob, and the main reason she claimed the division became profitable was growth in in-app advertising.

• GAM (DFP+AdX) was also a strong contributor to profitability.

• Google’s annual traffic acquisition costs (“TAC”) payment is $16 billion (2018?)

• YouTube makes up 21% of 2018 DV360 spend.

By power of elimination, if GAM and AdMob were very profitable this implies that DV360 and Google Ads may not have been as profitable. This further supports my argument that Ads was heavily subsidizing AdX with an artificially low 14% take rate, instead of its 32% rate on other supply sources.

Google witness: Mark Israel, Compass Lexicon

Israel is an economist and served as an expert witness for Google, primarily on market definition and to rebut the DOJ’s witness Robin Lee (see Day 9 coverage).

Imagine having a really smart economist uncle who knows nothing about advertising. Then imagine over the Thanksgiving table he mansplains how ad tech works to you. That’s what Israel’s testimony was like. I will try to be fair, but so much of his testimony was absurd.

Market definition

Israel’s primary argument is that the DOJ’s definition of three separate markets (ad serving, exchange, ad network) is “narrow and incomplete” and, in fact, the entire market should be considered as one. He criticizes the DOJ’s market as a) not including other media types, like video; b) not allowing for other methods of buying, like direct; and c) not considering alternative forms of ad tech.

Israel backed up these arguments with many slides and diagrams showing things like:

• The amount of time consumers spend on the web declined from 74% in 2010 to only 23% in 2022 (eMarketer).

• Ad spend in display declined from 70% of digital in 2012 to only 10% in 2022.

• Google share of the “display” market is only 10% according to eMarketer — which includes all social display in this definition.

• etc…

The thing that made Israel’s testimony absurd was that while he made some very solid arguments as to why Google Ads (on the buy-side) might not be a monopoly, he then irrationally extended those arguments to DFP and AdX (on the sell-side). Regarding AdX and DFP he said the following about switching (paraphrased):

He said this about DFP:

When the subject of Supply Path Optimization (“SPO”) came up he started by explaining how The Trade Desk and Criteo might skip exchanges, then pointed out that Facebook is essentially a leader in SPO since they only buy their own inventory.

On header bidding, he made the point that it is a viable competitor to DFP, so therefore is evidence there is no monopoly. This is just so ironic given that header bidding was specifically designed to break the alleged monopoly.

DFP

Israel went through each of the alleged monopolies and made his criticisms of Dr. Lee and of the market definition in general. The Dr. Lee criticisms were at times economically technical, so they may not be entirely elucidated below.

Here’s a summary of his DFP perspective:

• DFP prices have been steadily declining, which is not a sign of a monopoly.

• 86.3% of DFP publishers do not pay any ad serving fees.

• If publishers think DFP is being abusive then can just migrate their business to the app market where there is more ad serving competition (yes, he said this). He also indicated he thinks apps monetize better than web (ha!).

• Publishers can migrate to a subscription model.

• Since very large companies like Reddit and Amazon have migrated to their own home-grown ad server, this keeps DFP competitive even though most publishers could never to this themselves.

AdX

• Prices haven’t gone up over time, they have been 20% since DoubleClick days. If it was a monopoly they would have raised prices.

• AdX’s share has been declining. He bases this on a definition of “ad exchange” as any system that auctions ads, including all the social networks.

• 67% of DFP customers use four or more exchanges.

Google Ads

• The “Advertiser Ad Network” segment defined by the DOJ should include all the DSPs as well since they are buying platforms.

• While Google Ads has millions of buyers, 76.2% of revenue comes from advertisers that spend over $1 million/year, implying that these are the same customers as DV360.

• While Ads may offer CPC pricing, so does Amazon! And also, most DSPs have optimization strategies that can optimize to CPC, so it is the same thing.

• The percentage of campaigns on Google Ads that use CPC strategies have been declining in favor of ROAS and other goals, therefore this is exactly the same as DSPs!

Full stack

Israel made a strong argument that there are benefits that come from Google having a full stack across buy- and sell-sides:

• Not having two companies both maximizing profits, for example.

• Shared investment across both products

• Easier to operate

Lower take rates

In conjunction with the full stack argument, Israel repeated some of the same line of thought from yesterday’s witness regarding evaluating the combined take rate. Israel showed some math, which basically said that if you assume a $2.00 CPM, then:

Total Google take rate ≅ 1% DFP + 20% Adx + 14% Ads = 35%

He also showed data that this rate has been very consistent over time.

High investment levels

We got to see some data on the costs of all the engineers working on display at Google. Not a lot of independent ad tech companies investing $1.5 billion per year in salaries!

Cross examination

I left for lunch and got back to the courtroom a little late to hear an argument that went something like “Rotisserie Chicken is inconsistent with monopoly power.” Apparently there were some Costco metaphors let loose.

In any event, the DOJ indeed held Israel over a hot fire while giving him a spin. There was a magical exhibit labeled Demonstrative AG which showed four graphs:

• Demand for oil skyrocketing during the time of Standard Oil’s dominance;

• Demand for telecom skyrocketing during the time of AT&T dominance;

• Demand for PCs skyrocketing during the time of Microsoft’s dominance;

• And the chart the witness had shown with skyrocketing demand for display ads.

Removing Ads demand

The witness had conducted a simulation which showed that the removal of Google Ads demand would not actually hurt publishers that much since the same advertisers were available from other bidding systems. I didn’t fully understand the chart, but it seemed to indicate that only 5.8% of demand was really unique in Google Ads, and the rest could be reclaimed.

On cross, the DOJ lawyer tried to bring this very academic analysis back to reality, asking whether a small regional newspaper would expect to immediately get their demand back after turning off AdX.

AdX Direct

If you are not aware, it is possible to get AdX into non-DFP ad servers using a tag, known as “AdX Direct.” The problem with the product is that it doesn’t provide a bid, it just serves an ad. This makes it impossible to manage yield with this demand — in effect it forces you back to using a waterfall. According to the DOJ, only 1% of AdX revenue comes from this product.

The witness stood firm that this product gave publishers “exactly what AdX gives them,” which is absurd. Upon questioning he modified his statement to only what AdX gave publishers before unified first-price auctions, which might be technically true but also is absurd (because AdX already knew the floor price, whereas an external ad server would not).

Facebook

In a document from Facebook about growth of their Audience Network, the executives lament “there is always ad tech between us and supply.” Meaning Facebook realizes they have to go through DFP to get to the inventory. The witness intentionally misunderstood this and responded “given that Facebook has its own ad tech I cannot understand this.” This answer is convoluted because the witness had already put his stake in the ground that there’s no such thing as an “ad tech market”, so when asked something tactical about ad tech he answers in riddles.

Our first Larry appearance

I’ve been waiting for any emails from L&S and we got one on almost the last day. In 2008 Neal Mohan and Scott Spencer did a presentation on AdX and got this feedback from Eileen Naughton:

What’s next?

Tomorrow should be the first day of the plaintiff’s rebuttal. This may complete tomorrow or may slide into Monday.