Epic v. Google: everything we’re learning live in Fortnite court

When Fortnite was booted off the iPhone, its revenue tanked, we just saw in a graph— but after Google removed it from Google Play, we see its Android monthly revenue went from roughly $1M to $2.2M to a range of more like $0.5M to $1.75M during peaks. (I say "roughly" and "like" because the data points weren't labeled but appeared between lines that represented every half million in revenue).

That was measured between April 2020 and February 2021, according to the graph.

We also saw a slide suggesting most Fortnite players bought their V-Bucks elsewhere: between April and August 2020, only 2.8 percent of US users purchased any at all from Google Play, with 40.3 percent opting for other billing systems and 56.8 percent purchasing none at all.

• 99.5 percent of Android app developer revenue comes from purchases that happen after an app has been downloaded — so Google only makes its money back on distribution once a purchase happens

• Other platforms require their own payment system, like Airbnb, Amazon, eBay, Lyft, Uber, Taskrabbit, Roku, Steam, and Walmart

• We saw a slide with 21 different bullet pointed items Google has added to Play to support developers over the years

• In China, where Google is not so active, Gentzkow says app developers face a “fractured app store landscape” with service fees often higher than 50 percent (let alone 30 percent), and users see more fake apps, malware, and have more difficulty finding what they’re looking for

• Only 0.4 percent of the devices capable of running Fortnite (as of November 16th, 2021) were on the RSA 3.0 Premier Tier, and so “RSA 3.0 did not block Epic distribution”.

That’s Gentzkow, pointing out that Epic doesn’t seem to be biting the hand that feeds — it paid $471M to Sony, $260M to Microsoft, and $166M to Nintendo just from January to October of 2020, all platforms that impose a 30 percent fee.

According to the slide, it also paid $60M to Apple, a company it did sue — and $17M to Gearbox during the same period.

While we saw yesterday that big China phonemakers are choosing Google’s RSA 3.0 “Premier Tier” — which gives them more money from Google in the form of shared revenue, but also requires more compliance in return — Google’s economist suggests most phones have not gone that way.

Outside of China, only 7.4 percent of active Android devices are on the Premier Tier, according to one of the slides Google just presented, and roughly 27 percent of Android activations are on the Premier Tier as of July 2022.

(The latter does seem like a lot to me, though, particularly if you buy one of Google’s common arguments in this trial that it serves billions of people and so even small percentages add up to a lot.)

“Revenue sharing is a way of cutting the price by offering more back to the other party,” says Gentzkow, suggesting that Google’s special deals (presumably with OEMs, since that’s where Google was sharing a cut of Play revenue), were just another way to compete.

If he was referring to OEMs, I don’t quite follow the logic: who was Google competing with there? Google’s not competing to put Android on those phones, since OEMs couldn’t decide to put iOS on a phone instead.

Is Google Play the product that it’s discounting beyond free to pay those developers to carry? And if so, doesn’t that cut against Google’s argument that those developers needed Google Play and other Google apps to compete with the iPhone?

A pie chart from the earliest days of Android — or perhaps before it launched:

Symbian had 49.8 percent of the market, and BlackBerry had 15.9 percent, with 12.9 percent for Apple, 11.1 percent for Microsoft, 7.2 percent for Linux, and 2.1 percent for Palm.

I didn’t quite catch the full point that Gentzkow was trying to make, apologies. Google did point out that all those players have disappeared, save Apple.

We’ve moved on to an argument that Google’s MADA agreements are good for competition because they help Google compete with the iPhone, which helps sell phones, and in turn, helps developer sell more apps. “The more phones are sold the more things are going to be downloaded,” he says.

Continuing the “Macy’s and Nike” analogy where “Macy’s” is simply giving “Nike” a better deal in order to compete with “Galaxy Shoes,” Gentzkow says it’s natural that “Nike” wouldn’t be as likely to open its own store if it’s getting a good deal from “Macy’s” — that’s competition working, he suggests.

“Project Hug offers valuable things to developers in exchange for putting their apps into the store, and it’s natural” to expect the “Nikes” of the world to put their latest and greatest apps on Google Play right away as part of said deal, he argues.

Gentzkow suggests all the things that Google offered developers in its special Project Hug deals were simply the equivalent of giving them a “17% price cut,” and bid us imagine a Macy’s competing with a “Galaxy Shoes” for Nike’s business.

“Macy’s needs to compete for that business by offering Nike a great deal,” he says.

We’re now off for lunch.

“This was like a survey of users,” says Gentzkow, as Google presents a slide suggesting that 65 percent of Samsung Galaxy phone users have tried the Samsung Galaxy Store.

I wonder if Epic will challenge the source of that data and what it tells us. Yes, the Galaxy Store is right on the homescreen, so I suspect some users interacted with the icon, but a single user survey is not great data, and I wonder if the survey asked what they actually did with the store — and why.

(I recently bought a Samsung phone, and the first time I was pushed to interact with the Galaxy Store was when it sent a notification asking if I wanted to update a preinstalled Samsung app. I said yes.)

I feel better about Gentzkow’s arguments that Samsung and Amazon didn’t really produce compelling, consistent experiences and that Google had a substantial first-mover advantage with Play.

“Zero,” says Gentzkow — compared to 68 percent of Android phones in the US and 65.9 percent worldwide as of July 2021.

He says that, according to Google data, there were 3.2 billion monthly app installations outside of Google Play, 189M of them in the US as of May 2021 — and that those numbers went up 58 percent and 88 percent, respectively, compared to two years prior (July 2019, to be specific).

“It’s absolutely true that the number that happens through Google Play each month is quite a bit bigger... but I’m asking has this channel been blocked, do users not have the opportunity to use it,” he says.

“This level [of off-Play installs] shows lots and lots of people go out every month and are able to do this on their Android phones.”

We’re now hearing from Matthew Gentzkow, a Stanford professor of economics who also spent many years at the University of Chicago School of Business and says he publishes in all the top economics journals as well as Science and Nature. He also won an award.

He’s Google’s first economics expert and says he was tasked with answering two questions:

Did Google’s conduct create more value for consumers?

Did it block competition in a way that destroyed value for consumers?

His answers, as you’d expect: yes, and no.

He says that if one company is doing better than two other companies, it might simply mean it’s offering a better value — creating competition, says Gentzkow, not blocking it.

Earlier, we heard Google ask: “Square would charge more than Google would charge for that transaction, right?”

That’s because Google discovered that Square doesn’t have a special lower rate for microtransactions on digital platforms — developers wind up paying quite a bit for 99 cent transactions if they choose Square (not that they necessarily would).

But Epic just flipped that back onto Google with one incisive question:

“Does Square currently have any incentive for developing a schedule for microtransactions while there’s a tie in place that doesn’t allow it to provide microtransactions for digital goods?”

“Absolutely no incentive,” says Tadelis.

Google’s attorney suggested that eBay — which employed Prof. Tadelis at one point — needs to have an anti-steering provision because otherwise the buyer and seller could connect outside eBay’s platform and eBay would not get paid.

Tadelis agreed.

Out of the blue, Google’s attorney Kyle Mach asked this question: “You don’t think a case of this magnitude should be decided on back of the envelope calculations, do you?”

Incredibly, Tadelis seemed to know what he was talking about: “That is one of many pieces of evidence in my reports.”

I checked the live transcript, and this exchange seemed to be a complete non sequitur. What... were they talking about? It came right after a statement about how much Tadelis is getting paid.

Tadelis’ replies are getting amusingly snippy, by the way. Mach asked if Google “might have to charge fees for the other services it doesn’t charge for today.”

“It might not,” Tadelis replied. Then Mach and Tadelis went back and forth with “it might?” and “it might not.” several times.

Epic’s economist Steve Tadelis concedes that Google would get paid nothing to distribute apps if developers choose alternative payment processors — all other things staying the same.

(Historically and currently, Google only charges when a developer charges for an app or an in-app purchase, something it’s brought up repeatedly during the trial — it’s the “we only get paid when you get paid” argument, and I personally find it moderately compelling.)

Google’s making a host of other little points right now, alleging that Braintree is only available in a handful of countries, Square isn’t available in some big countries like Germany and India, PayPal fees vary across countries, and Tadelis gets paid $1,200 an hour plus a revenue share of his own. (Yes, another paid-to-be-here card played by Google, though Tadelis tells me afterwards he wasn't paid by Google to present his research, but rather to testify. I'll try to figure out the nuance later, but he's not contesting he was paid for his work.)

“Square would charge more than Google would charge for that transaction, right?” asks Google attorney Kyle Mach.

We’re looking at a chart from Tadelis’ report that shows that for a 99 cent in-app purchase, the second most common price set by developers, there are definitely worse options than Google.

Braintree and PayPal would eat 52.1 percent of that transaction, and Stripe and Square would take 33.2 percent — unless you went with one of their micropayment rates instead, which work out to low as 14.1 percent for a 99 cent transaction via PayPal and 10.1 percent for Stripe. (Tadelis admits he couldn’t find a micropayment rate for Square.)

At $4.99, the most common in-app purchase price, the highest effective rates of alternative payment processors are 13.3 for PayPal (or 6.8 for its micropayment fee) or 12.4 percent for Braintree. Others are quite a bit lower.

Google’s lawyer asks: “Had you included the Samsung Galaxy Store in this chart, it would look quite different, right?”

We’re now looking at Tadelis’ bar graph again — the one that unfavorably compared Google’s nearly 30 percent fee to the 6–10 percent fees of payment processors — but now with Apple and Samsung’s stores added. Those bars reach much higher than Stripe and PayPal’s, of course, making Google’s look more normal.

Google once again makes its point: stores aren’t just payment processors. They have additional value. Tadelis concedes they’re not exactly the same but says it depends on what you’re using them for.

We just saw an intern document where Google attempted to use game theory to figure out what would happen if it decoupled Google Play Billing from Google Play app distribution and decided to compete with other payment processors itself.

It included a graph that suggested many developers would change to a rival payment processor right away — “some large developers would take advantage of billing optionality no matter the price,” wrote Google — but that most others wouldn’t pick Google until or unless it offered a substantial discount.

Tadelis showed us that Google’s User Choice Billing, which nominally offers a mere 4 percent discount (though secretly goes far lower), wouldn’t come close to the place on the graph where it would convince developers to pick Google.

Tadelis says User Choice Billing is not a good deal, and Google knows it.

I wish I could show you a picture of the graph since it’s difficult to describe. No cameras in the courtroom; we should get a copy after the trial concludes.

It’s Google’s turn to question Tadelis now.

We’re looking at a demonstrative slide titled “Web Browser Payment Solutions Services Are Not Substitutes,” showing the many steps it can take to buy digital goods on the web instead of through an app.

He says internet companies know “very well” that any amount of friction in this payment process leads to serious drop-off and that he’s seen it firsthand: “Both at the times when I was at eBay and at Amazon, there were many projects that dealt directly with friction.”

Epic asks if app developers could use links in their apps to speed things up. Sure, says Tadelis, but Google doesn’t allow that. Epic puts Google’s anti-steering provision on-screen:

“In-app user interfaced flows, including account creation or sign-up flows, that lead users from an app to a payment method other than Google Play’s billing system as part of those flows.”

Anti-steering is the one place Epic won a tentative victory against Apple, BTW.

I haven’t seen enough of the docs myself in court to say for sure — only a taste — but Epic’s economist is telling us about Google documents that show the company feared its Play Billing would be displaced.

“Google was well aware that severing the tie would put price pressure on payment solutions they could provide,” he says. We saw a brief snippet of a document showing a multi-step process where other payment providers would rise up to fill a mandatory Play Billing vacuum.

He says another chunk of the document showed that Google saw benefits in “laddering up,” or using its power to become dominant in a subsequent area — like how Netflix’s DVD business became a launchpad for its streaming business, which became a launchpad for its own TV and movie productions, he says.

“What would you expect to happen to price to consumers if the tie were severed?” asks Epic’s lawyer, referring to a hypothetical world where Google didn’t require its 30 percent Play Billing fee.

Tadelis says that “savings from the reduced fees would naturally be passed onto the users.” As anyone who’s lived through inflation-as-excuse-for-price-gauging will tell you, that seems a little optimistic?

Judge Donato picked up on this instantly, pausing the testimony to ask a question of his own: “Let’s say in a hypothetical world, Google eliminated all fees. Why do you expect a user would see any benefit in that, in the price a user pays for an app?”

Tadelis suggested that if he opened a pancake shop and suddenly the cost of milk and eggs drop, he’d make more profit right away — but if he drops the price a bit, he can sell even more goods and make a wider margin.

“Macroeconomic analysis shows that some of that will optimally be passed on to consumers,” he says. The judge seems satisfied, at least from where I’m sitting.

“Most of the value of the charges happen at 30 percent, not a 15 percent.”

“If the developer that’s charged 30 percent is selling a million dollars of product and the developer charged 15 percent is selling $1 of product,” then the fee is closer to 30 percent on average, he says.

(In 2021, after Epic filed its lawsuit, Google reduced its cut to 15 percent for a developer’s first $1M in annual revenue. Some companies pay far less than 15 percent, though — even as low as zero.)

That’s the second argument from Epic’s second economist, and we’re looking at a slide listing basic “competitive harms”: foreclosure, high prices, and “innovation and quality.”

On the first point, foreclosure: “When Google doesn’t have to compete with other payment solution providers, they lack the incentive to offer the innovations that are needed,” says Tadelis.

On the second, high prices: “If the tie were severed, could they charge 15–30 percent for a payment solution? And the answer to that is no,” he says.

“Anyone who is profit maximizing would not choose to pay a 30 percent fee when they could pay a 9 or 10 or 6 percent fee,” he adds.

We’re looking at a slide showing Google’s average in-app purchase fee (29.4 percent as calculated by Tadelis) versus the 5.5 to 10 percent charged by PayPal, Stripe, Square, and Paddle.

Keep that in mind — he says he was assigned to look into Google’s behavior and concluded that Google “forces developers that wish to use the Google Play Store to use Google Play Billing for all in-app purchases of digital content within their apps (a ‘coercive tie’)” among other findings.

He explains coercive ties simply: if you want product A that I’m selling, you have to buy product B as well. But for it to be impactful, it also requires there to be no viable alternatives, he says. If Nike makes you buy a pair of socks with its sneakers, and you don’t like those socks, you could simply buy sneakers from Adidas instead.

We’re looking at an amusingly simple visual aid: a thick clip art of a physical rope with a prominent knot between Google Play and Google Play Billing.

So, Epic has two economists, including Doug Bernheim, whom we heard from yesterday; expect Google to have two as well.