Since the launch of the App Store in 2008, but especially since 2009-2010, Apple and Facebook have played fairly well-defined commercial roles in their respective relationships with mobile app developers: Apple as content platform and Facebook as advertising platform. But strategy revisions by both companies — driven by slowing smartphone sales growth due to high levels of smartphone penetration and a maturing yet highly lucrative app economy — have put them on a collision course.
Until recently, Facebook and Apple were in a sort of alliance of necessity, given their common enemy in Google: both companies were incentivized to not brazenly confront the other’s core business on mobile, since Google represented a more pressing and harrowing threat to both. But that tenuous solidarity may have been eroded by product and policy changes from both companies, creating a tension that has the potential to spill over into an all-out war.
In this article, I give an overview of the recent history of both companies’ evolved strategies related to the app economy and observe motivations of each for encroaching on the other’s commercial territory: Facebook into content and Apple into advertising and services. I end the article with some thoughts on how this situation may play out — whether it takes the form of a war or merely a tense relationship, it seems that the mobile ambitions and product trajectories of Facebook and Apple are poised for conflict.
App Stores within App Stores
In April 2013, Apple removed AppGratis, an “app discovery app”, from the App Store. The reasoning it provided for the removal was that the app violated its terms of service, specifically term number 2.25, which states:
2.25 Apps that display Apps other than your own for purchase or promotion in a manner similar to or confusing with the App Store will be rejected.
Apple’s terms of service have since been re-written and renumbered, but the spirit of 2.25 lives on in the form of term 3.2.2:
(i) Creating an interface for displaying third party apps, extensions, or plug-ins similar to the App Store or as a general-interest collection.
And so it was that Apple made a high-profile example of a company that had flouted its terms of service (AppGratis had raised more than $10MM in VC funding mere months before its removal from the App Store), making it clear to developers that the “app store within the App Store” model would not be tolerated.
And in the summer of 2014, Apple took this line of policy a step further: following the announcement of a set of new discovery features within the App Store at that year’s WWDC event (trending searches, an explore tab, etc.), Apple began rejecting apps that implemented rewarded video ads as a means of monetization.
For those that don’t know: rewarded video ads give users access to premium functionality (in games, usually virtual currency or extra lives) in exchange for watching a video ad. With this change in protocol — which was never explicitly articulated by Apple — Apple was expanding the scope of its interpretation of “displaying apps” from its terms of service to mean any form of showcasing another app, whether as an “app store” or otherwise.
Apple eventually reversed this stance in late June 2014 after severe backlash from developers — rewarded video ads, especially now, are a major source of revenue for many app developers — but the looser interpretation of this policy is still selectively enforced as it applies to egregiously incentivized installs, particularly for cross promotion purposes between a developer’s apps. Speaking broadly, Apple has no problem with developers using aggressive tactics such as pop-up interstitial ads to inform users of one app about another app: after all, being able to move users from app to app within a portfolio helps developers invest more into their products, especially from a marketing standpoint.
But Apple frowns upon coupling these ads with rewards tied to downloading another app: for example, in a game, if a pop-up interstitial ad launches as soon as the app loads and promises the user some amount of currency in exchange for downloading the developer’s new game. Apple still tends to reject apps that feature these types of ads — incentivized for a download, not opt in — in an egregious way (eg. launched as soon as the user opens the app).
What was (and is) the source of Apple’s antipathy to these types of app promotion schemes? The lamentations of “broken app discovery” (“app discovery” is used as a catch-all term to mean the way people find new apps) are ubiquitous, and these sources of installs have always been an effective means of pushing users. Traditional wisdom held that Apple frowned upon these forms of app discovery for two reasons:
- Steve Jobs’ infamous antipathy towards advertising (in early 2016, Apple shut down iAd, its own mobile advertising network);
- A maniacal focus on user experience to set the App Store apart from Google Play as a “premium” platform. Aggressive ads have the potential to debase the user experience, and Apple cared about hardware sales, not App Store revenues (which are generated as a 30% cut of all in-app purchases made and are meager as a proportion of total company revenues: in 2015, Apple generated about $6BN in revenues from the App Store via its 30% cut of in-app purchases versus $233BN in total revenue in 2015). To the extent that the App Store was strategic to Apple, it was as a differentiator that drove hardware sales, not in-app revenues.
As an extension of this traditional wisdom, Apple was seen to want to control distribution via featuring and its top charts to focus engagement into the apps it felt best represented and showcased its hardware. Each Thursday, the App Store homepage is refreshed to unveil that week’s darlings of the app economy: the apps that were selected by the App Store editorial team to be exhibited to the entire iPhone- and iPad-owning public as dignitaries of Apple’s digital fiefdom. The “Editor’s Choice” designation — the massive banner at the very top of the homepage — is the featuring accolade par excellence, delivering potentially millions of installs to recipients, of which there are only four or five each week out of thousands of app submissions.
Ads within the App Store platform — that is, from within the App Store in the form of featuring but also the ad units sold within apps — were regulated by Apple to ensure the primacy of the App Store experience and also to give Apple control over the levers that drove App Store notoriety. This created space outside of the App Store to build advertising platforms that could help app developers promote their apps without the limitations of those rigorous standards.
It is in this space that Facebook has flourished, with its mobile advertising business driving quarter after quarter of remarkable financial results.
Facebook first launched its mobile app install product in the summer of 2012 following what many outside observers deemed to be a “disappointing” IPO. The company’s stock price languished well below its debut price of $38 for about a year after its IPO, until July 19th, when leaks began emerging about Facebook’s impending quarterly results, which would reveal impressive revenue growth on the back of its new mobile app install ads product. Sure enough, Facebook’s July 25th earnings call revealed that mobile app install ads had taken the company to $1.8BN in revenue for the quarter from $1.2BN the year before, a 50% jump. Facebook’s stock rallied by 30% on the day of the release and precipitated a price climb that would continue, more or less unabated, to this day:
Facebook’s mobile ads were at first confined to its own news feed, but the company embarked on a series of initiatives after 2013 that would increase that scope:
- In November 2013, Facebook brought ads to Instagram, the photo sharing app it acquired in 2012 for what was then an eye-popping $1BN. These ads were mostly brand-oriented at first, but in the summer of 2015, Instagram launched mobile app installs ads in the model of those found on the Facebook News Feed;
- In January 2014, Facebook announced that it would begin to roll out its Audience Network advertising platform, which would be integrated into third party apps and allow them to serve ads based on Facebook’s proprietary user data and targeting tools;
- In February 2014, Facebook announced that it had acquired WhatsApp for an Instagram-belittling $19BN. Little reported in this transaction was the fact that its acquisition of WhatsApp gave Facebook the phone numbers of 450MM people — those phone numbers could eventually be paired with Facebook’s own proprietary data to create pronounced targetability for advertisers on the basis of the kind of data that traditional marketers typically hold on users. In the summer of 2016, Facebook eventually announced that it was doing just that: pairing WhatsApp phone numbers with Facebook profile data;
- In March 2014, Facebook rolled out its News Feed video ad product, which allows advertisers to show video ads in users’ News Feeds that start automatically upon being scrolled over. Video has been the biggest mobile advertising story of the last three years, and Facebook was one of the first large advertising platforms to integrate it natively into its product in a meaningful way;
- In March 2015, Facebook announced that it would expand its Messenger app beyond mere messaging, creating something similar to the OTT messaging platforms that are popular in Japan. Messenger had been “spun out” of the Facebook app in the summer of 2014 with a forced install, rendering it one of the largest apps in the app economy.
The net result of the above is that, by late 2016, Facebook owned three of the most-used apps in the app economy as measured by Nielsen: Facebook (#1), Facebook Messenger (#2), and Instagram (#3), in addition to WhatsApp:
Facebook and Instagram are mobile advertising juggernauts that have delivered massive amounts of enterprise value to Facebook on the backs of their mobile app install ad products.
But the speed of Facebook’s ascent introduces a rarefied problem: it’s running out of people on the planet to bring onto its platform, to such an extent that it’s even trying to deliver free wireless internet to rural areas in India just to onboard more consumers of ads. Since Facebook can’t will into being more consumers of ads, it has to show lots of ads to its existing user base. Historically, Facebook’s CFO, David Wehner, has been charged with addressing the question of “ad load” — that is, the density of ads in Facebook’s proprietary content streams, like the News Feed on Facebook and the image feed in Instagram — in the company’s earnings calls. His mostly canned pronouncements on ad load are usually benign enough: vague cautions about a future time.
But in last quarter’s earnings call, held on November 2nd, Wehner’s language changed slightly but deliberately when he addressed ad load. Specifically, his proclamation had changed by one word: from revenue growth being impacted “accordingly” (as he said for Q2 2016) in the future by ad load saturation to “meaningfully”. This one word caused an 8% drop in the company’s share price in after hours trading (the company’s quarter had been phenomenal: 5% growth in MAU, reaching more than 1BN mobile users).
Facebook can only grow its advertising revenues a few different ways. The first is ad load, which, as Wehner indicated, is more or less saturated on existing properties (namely, News Feed and Instagram).
The second is user base growth, which is strong overall for the company but concentrated mostly in the developing world: for the two years ending Q2 2016, Facebook’s MAU only grew by about 1 – 1.5% each quarter in the US and Canada (Asia Pacific and the nebulous “Rest of World” regions are where Facebook is really exploding).
The third is ad inventory price (usually related in terms of CPM, or cost per thousand impressions), which tends to be volatile quarter-to-quarter but has been more or less flat over a longer timeline (especially as Facebook started driving installs via its Audience Network, which runs at a lower price point than its premium native inventory.).
That said, Facebook is going all-in on video — the company has been aggressive about implementing video content natively into the news feed and orienting many of its advertising products and tools around mobile video. Mark Zuckerberg told Buzzfeed in an interview that he wouldn’t be surprised if “most of the content that people see on Facebook” in five years’ time is in video format. Video is clearly an important — possibly the most important — content initiative for Facebook, and it’s also the best performing ad format on mobile. If Facebook is able to consistently roll out new editing tools (such as its auto captioning tool), new analytics features (such as its Nielsen reports for branded premium inventory, which I discuss in my 2016 GDC presentation about the launch of Angry Birds 2), and new video ad formats (such as the vertical ad unit that launched on Instagram in November 2016), then Facebook could conceivably drive prices for its inventory up over an extended timeline.
The fourth potential source of growth is what I call “inventory surface area”. In order to serve more ad units without making News Feed look like a handheld version of Times Square (and the attendant perception of tackiness such a sight invokes), Facebook extended its ad reach into other developers’ apps via Audience Network. If the Facebook and Instagram apps were giant zeppelins bedecked with big LCD advertising screens on the sides, then other apps were much smaller blimps pulling banners for Joe’s Plumbing. With the Audience Network, Facebook allowed those smaller blimps to plug into its vast data store on users to serve ads with more sophisticated targeting, with Facebook acting as a broker and taking a cut of each ad view.
But what if Facebook could subsume those apps, bringing more content onto its platform and thus increasing the “surface area” of its platform, allowing it to serve ads without increasing the density of ad load? In May 2015, Facebook launched a beta of a program called “Instant Articles”, which allowed publishers (mostly news outlets) to host their content directly within the Facebook mobile app, meaning it’d open immediately upon being clicked and not require the user to leave the Facebook app. The program was a good deal for publishers: it made their content available to the largest audience on mobile in a convenient way, and it also alleviated the need for publishers to “figure out mobile”, as Facebook was so generously offering to sell ads on that content for a 30% cut of the ad revenue. Facebook launched the program globally in April of 2016, with some of the largest news publishers in the world as participants.
If the Audience Network syndicated ads onto other blimps, Instant Articles was the process of cutting those other blimps up and sewing their material into the Facebook zeppelin. The “surface area” of the zeppelin increased, and any ads served on the canvas corpses of those other content providers didn’t increase the density of ads in the Facebook user experience. Instant Articles allows Facebook to serve third party content in its own app, monetize that experience with ads, and not increase ad load.
This content absorption strategy can be seen everywhere on Facebook and Instagram, from Instant Articles to Facebook Live Video (Facebook recently launched mid-roll ads in the Live Video broadcasts users produce from its app), to Instagram’s product identification feature (Instagram identifies retail products in pictures and provides a link for purchasing the item in question, with plans to facilitate that transaction in the Instagram app soon). But the strategy is perhaps most conspicuous in its new Instant Games product.
In late November 2016, Facebook launched Instant Games, which is essentially an app store for HTML5 games located within the Messenger app. These games are loaded “instantly” (just a few seconds) and can be played socially; featured in the inauguration of the platform were games from some of the biggest mobile game developers (King, Zynga, Game Insight, Bandai Namco) whose products would otherwise be playable via an app download. Information about how Facebook will eventually monetize Instant Games is light, but if it uses the Instant Articles playbook, it means that Facebook will take a 30% cut of advertising revenues from ads that it serves, and will likely do the same with in-app purchases revenues should it ever allow those on the platform.
Instant Games represents a major coup for Facebook: content that might otherwise exist on the App Store now exists on Facebook, and Facebook will potentially take a 30% cut of advertising revenue generated in that content without users ever leaving Messenger. And this whole process starts from within something that looks conspicuously like an app store.
A Shift to Services
In December 2015, Apple announced that Phil Schiller, its head of worldwide marketing, would take ownership of its App Store ecosystem across all of its platforms. Within a few months, Schiller’s promotion had a massive impact on the App Store and Apple’s perceived attitude towards it. In a pre-WWDC interview (which was almost unthinkable in its own right) with The Verge, Schiller laid out a new vision for the App Store, with an emphasis on subscription revenues and, astonishingly, a new App Store search ad product that would allow developers to bid on promoted placements in search results.
It’s impossible to overstate how colossal of a change this represented, both in terms of policy and philosophy. According to the traditional popular wisdom outline above, Apple was indifferent to app revenues: the App Store was just a means of keeping people attached to their iPhones. Rumors had swirled about the possibility of App Store search ads ahead of WWDC, but they were easy to dismiss: I wrote an article titled “Putting App Store Revenues into Context” just before WWDC explaining why I thought that Apple would never abase itself by injecting ads into its ecosystem, control over which it exerted through the content stewardship of app featuring.
I was wrong, and Apple did introduce search ads to the App Store, allowing developers to promote their apps alongside the results for searches like “hotel booking app” and “photo editing” but also for terms related to competitors’ IP such as “clash of clans” or “angry birds”. In terms of the app ecosystem, search ads probably had a net positive impact on developers: they exist as one more means of reaching users in a crowded marketplace, and since they are shown at the “last mile” of the app download process (in the App Store, when a user is primed to download), they generally perform better than other ads. Apple revealed in early January of this year that Search Ads convert — that is, produce a download from a click — at above 50%, which is quite high by mobile advertising standards.
But why would Apple change course on App Store revenues and advertising? One reason is that hardware sales were falling as a result of high smartphone penetration: Apple posted a decrease in year-on-year revenue for the first time ever in Q2 2016, when every single one of its hardware businesses experienced declines. Smartphone sales growth has slowed worldwide: mobile phone penetration in the US and the biggest EU countries is above 90%, and 65% of Americans already own a smartphone. Smartphone sales are generally driven by replacement now, and the replacement cycle for the iPhone has increased for the iPhone over the past few model releases to three years.
The second reason is probably that Apple saw how much money Facebook et al were making on mobile app install ads and felt entitled to a cut of it. Facebook wasn’t the only player benefiting from the heady mobile advertising days of 2012 through 2016: big businesses were built on the process of impelling people to download apps. Applovin, a San Francisco-based mobile ad network that was founded in 2012, was acquired by a Chinese private equity firm in late 2016 for $1.4BN. More than one billion dollars in equity value was built for just one company in less than five years by peddling clicks and installs, and the Applovin story (although exceptional) is not unique. Apple was making other companies rich, and even though those numbers were a pittance compared to Apple’s annual revenues, the money flowing elsewhere was still green.
And so it was that in September 2016, via the iOS 10 operating system for iPhone and iPad that had been announced at WWDC alongside search ads, Apple introduced another change: IDFA zeroing. An IDFA is the unique advertising identifier that is attached to each iOS device, having replaced the UDID, which was a unique device ID that Apple deprecated in 2012 over privacy concerns. When the IDFA was introduced in 2012, a function called “Limit Ad Tracking” (LAT) was added to the privacy settings of iOS that allows users to indicate that they don’t want their IDFA used for ad targeting purposes. But even with LAT activated, apps had access to users’ IDFAs: they were merely compelled by the iTunes Connect Terms & Services to not use it for targeting, almost like an honor system. While Apple sporadically rejected apps that accessed users’ IDFAs and didn’t show ads, the LAT setting as implemented in 2012 didn’t really present a problem for most mobile marketers.
Before continuing to the 2016 implementation of LAT, it’s important to explain why the IDFA is relevant to advertisers. When an advertiser serves an ad on the desktop web, they can use the URL that the advertisement forwards to to capture a lot of information about that ad click: where the ad was served, from which campaign it was served, etc. This information is encoded in variables in the URL behind a question mark: you’ll often see strings like “?source=X&campaign_ID=Y” at the end of URLs. All of this data can be stored when the user reaches the click destination of the ad by the advertiser to help them attribute those clicks to various campaigns and calculate the return on investment of their marketing spend.
That information can’t be encoded for mobile app install ad campaigns — or, at least, the platform operators won’t let it. So when a developer’s app is opened by a user, the developer has no background information on the provenance of that install. To alleviate this problem, an entire industry was born: mobile ad attribution.
As a gross oversimplification of the process of mobile install attribution: a developer routes app install ad campaign clicks to a third party “attribution” service, which captures a lot of information about the user that clicked on an ad (including, if they’re using an iPhone or iPad, their device’s IDFA) before forwarding the user to the appropriate page in the App Store or Google Play. The user either downloads the app or doesn’t; if they do, when they open the app, that attribution service transmits, via an SDK that is implemented into the app, that same information back to its server. The server compares the data on the click and the install and determines whether the person that clicked the ad is the same person using the app now. If a match is made, that user is attributed to the click that was recorded and the campaign that click was generated by, and the developer can then estimate the expected revenue those users will generate over time (the users’ Lifetime Customer Values, or LTVs) and calculate an ROI on those campaigns based on their costs.
But with iOS 10, for users with LAT turned on, Apple was changing its policy: rather than leave the IDFA accessible, Apple would return a string of 0s, meaning it’d be impossible to tell one LAT-activated user apart from another when tracking marketing clicks and installs. This was a massive policy change, and it went more or less unreported: between 10-20% of all iOS users have LAT turned on, meaning targeting those users (there’s no anchor for targeting data) and tracking their installs (they all look the same) via IDFA is impossible.
Why is this important? Because Apple doesn’t need an IDFA to serve relevant ads in search: marketers have all the targeting they need from the intent baked into search keywords. So in iOS 10, Apple introduced a proprietary ad product that performs well independently of ad targeting while simultaneously limiting the ability of other networks to serve targeted ads.
Of course, networks and attribution platforms found a way mostly around this — building “fingerprints” of users to serve as unique identifiers — but that’s beside the point. The IDFA zeroing episode highlighted just how much power Apple has to shoot massive shockwaves throughout the mobile advertising ecosystem and send advertisers, ad networks, and technology companies alike scrambling to react. Apple controls the point of distribution, and so in its poker game with participants in the app economy, it holds, if not all the cards, at least the winning hand.
The Golden Fleece
Atreus and Thyestes were two brothers who feuded bitterly. Atreus became the King of Mycenae and vowed to sacrifice his best lamb to the god Artemis, but when he noticed that his best lamb had a golden fleece, he killed the lamb and gave the fleece to his wife for safekeeping. Unfortunately for him, Atreus’ wife had started an affair with Thyestes, who surreptitiously stole the golden fleece and then suggested publicly that whoever owned it should be king. Atreus agreed, and when Thyestes produced the fleece, Atreus forfeited his crown.
If the app economy is Mycenae, and the golden fleece is the ability of a company to direct app installs and control users’ attention, then who owns it? From 2008 – 2012, before the ascendency of the freemium business model in mobile and the massive economy that it precipitated, it was Apple: Apple was the mobile kingmaker and determined who installed what. But in its rise, Facebook wrestled the golden fleece away from Apple — it became the distribution gateway for the mobile app economy, directing installs on the basis of developers’ marketing spend. Apple has shown a willingness to claim the fleece again through its iOS policy changes and the introduction of search ads: if it can build a massive wall around its content platform, it can once again control who installs what. And Facebook, by subsuming third-party content in the form of articles, games, and who knows what else in the future, could be establishing a closed-loop content platform that eventually eats up more and more of users’ time on mobile to the exclusion of other apps.
Whether the two companies enter into a cold war, a hot war, or peacefully co-exist, it’s not hard to imagine any number of actions that both companies could take, either in the spirit of being downright hostile or merely passive aggressive:
- What if Apple refuses to allow developers to link directly to their App Store pages in the future, instead only allowing them to link to a search result for the app’s name? That’d result in Apple being able to siphon off some amount of advertising revenue on every ad click by positioning its own ads first (passive aggressive);
- What if Apple cuts off linking to the App Store altogether or somehow blocks the ability for mobile app installs to be attributed? (hostile);
- What if Facebook starts allowing for developers to buy ads that link to their Instant Games content? (passive aggressive);
- What if Facebook builds out an IAP economy for Instant Games and, to add insult to industry, encourages users to purchase credits on the Facebook website so as to avoid paying Apple’s 30% platform fee, as Spotify does? (hostile)
However far fetched, these are but a few potential actions either side could take that might cause simmering animosity to bubble over and lead to a Mobile Missile Crisis. Much of this depends on whether Apple decides to enforce its “app store within the App Store” policy as a line in the sand: if it gives Facebook a pass on Instant Games, who knows what comes next, and from where.
Because if you squint and stare hard enough at various parts of the app economy, other “app stores within the App Store” appear. Google recently began allowing users to order and track Uber rides from within the Google Maps app. And Facebook’s News Feed has looked like an App Store for a long time, albeit one in which a download is replaced with a scroll: integrated news feed videos, the forced download of Messenger, Instant Articles, the Marketplace tab, etc.
One point that bears making is that, despite the bizarre hysteria claiming otherwise, the app economy is in the midst of a multi-year boom: Apple paid $20BN to developers in 2016 (taking about $8.5BN for itself in platform fees), and the App Store set a new sales record on New Year’s Day 2017, generating $240MM in revenues. The big players in the app economy are actually huge, commanding massive audiences, generating extraordinary amounts of revenue, and propelling astonishing growth in the mobile marketing industry.
The contests for territory on mobile — by Apple and Facebook specifically as the subject of this article, but also by Google and others — underscore an important point: at enough scale, everything can become a platform, like a Jenga tower reaching for the sky. The question is: how far down the layers of the tower do you have to go until you reach true terra firma, and is Apple willing to enforce itself as that. Perhaps Apple and Facebook both decide they’re better served as frenemies and neither lets its animosity or ambition manifest into overt opposition to the other.
It’s worth noting that Atreus was ultimately killed by Thyestes’ son, whom Thyestes conceived with his own daughter.