Questions about a Google Break-Up

Questions about a Google Break-Up

By now I’m sure all of this newsletter’s readers are aware that Google lost its ad tech antitrust case, and was found to be an illegal monopoly in both the publisher ad serving market and the “open web” ad exchange market. Despite Google’s ridiculous and tired spin, this was a muscular victory for the DOJ.

Of course, Google will appeal. And there’s the possibility this judgment gets horse traded with the search case in some way to save the Chrome browser. And there’s the unpredictable Trump card, where a very large wire transfer can make a lot of legal problems go away.

With all that said, I’ve long said that I believe Google would be fine with a spin of some or all of its “Network” division, which Eric Seufert accurately refers to as “vestigal.” Let’s spend some time gaming out a spin-out and asking some questions.

What gets spun?

There seems to be a lot of confusion on this subject, but I think it is important to look at the facts. The DOJ, the EU, and Canada have all asked for the same thing: The spin out of GAM and AdX. Look, the bureaucrats in the EU even created this highly detailed graphic to explain what they want:

Absent a grand bargain to save Chrome, etc there is no hard evidence of any kind that DV360, Google Ads, Google Analytics, or Google Campaign Manager (aka DFA) are on the table as potential spins. There’s lots of wishful thinking, but no evidence.

What about AdSense?

We’ve hit our first speed bump in our neat little spin-out plan. AdSense is the largely self-service publisher ad network that monetizes remnant inventory, and it predates GAM and AdX. It has also largely been omitted from any of the legal discussions, so one might assume it is simply not an issue.

Maybe. The problem is AdSense is deeply tied with both GAM and AdX. The free version of GAM (aka “small business,” a name they have scrubbed from the internet) is essentially an AdSense on-ramp; without AdSense that product has no reason to exist. AdX, too, generates a great deal of the demand on AdSense, so its separation will require a wholesale reworking of yield for publishers on that platform.

Post spin AdSense goes back to its roots as a remnant solution for publishers, with demand almost entirely coming from Google Ads.

Is AdMob part of GAM, AdSense, both, or neither?

AdMob seems like it exists in a parallel universe, immune from the legal issues in open web, yet enjoying some of the same benefits of deep integrations. If you’re not familiar with AdMob (which is totally understandable given how little Google talks about it), the mobile in-app business includes an ad server sometimes called “AdMob ad platform” and an ad network, that competes with $META ( ▲ 0.44% ) , $APP ( ▼ 0.44% ) , $IS ( ▼ 2.83% ) and others.

While we don’t expect AdMob to be subject to a spin, the ad serving offering for app publishers is basically GAM. So in all likelihood, AdMob will be forced to stop offering an ad server as part of its package. The SDK is another issue. While Google tries to avoid saying this in its documentation, AdMob and GAM share an SDK, called the Mobile Ads SDK.

Post spin AdMob loses some of its appeal to app publishers, as they have to get their ad server and monetization solutions from different places.

How will Google Ads demand flow?

In addition to the spin, the court should be expected to impose various behavioral remedies to prevent the re-emergence of anti-competitive behavior. The biggest question is going to be around how to regulate the demand coming from Google Ads and DV360.

Imagine if the day after a spin of AdX Google announced an exclusive partnership with say, Pubmatic, to direct all or most of its demand to that exchange. The monopolistic effect would be recreated and publishers would be little better off than they are now, though without the ad server tie.

To prevent this theoretical power grab, the court might try to enforce some kind of fair dealing standard that forces Google to bid evenly across supply sources. Sounds easy. What factors would Google be allowed, by the court, to use to determine fairness?

• Take rate

• Presence of fraud/bots

• Data transparency

• Outcomes

• Data rights

• etc

These are the exact same parameters that every DSP and ad network use to evaluate inventory, typically in a very dynamic and optimized feedback loop. How could the court possibly determine whether this was being done fairly? What if post-spin, 50% or more of Google’s demand continued to flow to the newly-independent AdX because the algorithms continues to see the benefit of that inventory? That was basically Google’s defense in court — that AdX has less fraud, better prices, and more data — so naturally it gets the bulk of the demand.

This also brings up the hairy issue of how a post-spin Google will evaluate its O&O properties versus open exchange. We already know that DV360 and PMax prefer YouTube, AdX, and AdSense to the open web. If you remove AdX from the options, is the court going to enforce a proportional redistribution of that demand to the open web?

Further, what if Google were to simply stop bidding on open web from Google Ads entirely? Would the court allow that?

What are the behavioral lines around AdSense and AdMob?

We’ve already discussed the grey areas between Google’s sell-side products, but post-spin the remaining parts become the subject of court oversight. To give an obvious example, imagine that after an ad server spin-out the teams supporting the remaining parts of AdMob or AdSense decide to support open web ad serving — that certainly would not be allowed.

If we assume that both AdSense and AdMob continue to have a demand advantage from Google Ads, they could pretty easily use that leverage to recreate some of the anticompetitive power. Here’s a quick list of things that may need to be prohibited:

• Offering an ad server

• Allowing other bidders to bid into AdSense (thus becoming an exchange)

• Allowing AdSense to optimize other tags

• Making either Ads or DV360 demand exclusive

• etc

What about the spun companies?

The spun companies would still be hugely important, though their financial outlooks would be substantially dimmed.

From court filings we know AdX has ~50% market share and GAM has ~90% share. AdX’s share has been gradually declining since the emergence of header bidding and the company’s gradual removal of its many advantages like “last look,” etc. If kept as a combined entity, AdX would still have the advantages inherent in being part of the ad server: shared workflow, shared identity, common infra and low latency, among others. But it would gradually lose all of those same advantages as the Google buy-side spreads its demand elsewhere, and overall publisher yield through AdX would certainly decline.

You may have noticed the plural “companies” in this header. Everyone seems to be assuming that GAM and AdX would be spun into a giant new public company, call it “DoubleClick,” maybe. But there’s also the possibility they are spun separately into two companies, which would make the calculus much more complicated. What if AdX was spun separately?

Without AdX, the newly spun GAM would be much less financially attractive. Ad serving rates hover around a cent per thousand, and the total SaaS revenue from GAM are probably ~$500 million per year. Not bad, but its got a monopoly (the court said so!), so where exactly is growth coming from? In this separate spin-out scenario the new GAM would certainly be prohibited from buying or operating an exchange, ad network, or other monetization method. Boxing the company in this way would lend itself to the idea floated of making it a B-Corp or some other form of non-profit.

AdX without the ad server and without preferential demand from Google would also be a substantially less valuable entity. Market share would decline because of the Google ads demand, and products like programmatic guaranteed and open bidding would have to be rethought from scratch. Also, presumably, this entity would be prevented by the court from operating on the buy side or hooking up with a competitive ad server.

Readers may be asking why the court would force a value-destroying spin-out like this. But that’s a fundamental misunderstanding of the court’s goals. The point is to stop anticompetitive behavior, not to preserve shareholder value.

When will we get answers?

We don’t know the timing for the remedies phase of the trial, but you can be pretty sure we’ll be covering it here.

Reading list

• Facebook antitrust, Zuck on the stand (link)

• OpenAI developing social network? (link)

• Two lawsuits allege the Trade Desk secretly violates consumer privacy laws (AdWeek)

• Jimmy Fallon and Bozoma Saint John will lead a marketing agency staffed with contestants on a new NBC competition series called “On Brand.” [Variety]

• What was the Infosum acquisition price? Digiday says $150 million, but Reuters says $63 million (link)