Meta’s $10 billion problem: Slowing ad revenue could hurt its ambitious metaverse plans. Here’s how the company hopes to rebound

Because this is an earnings call, Zuck also wants to make sure the analysts understand that the metaverse is about making money for Meta. “Our other focus for Horizon is building out the metaverse economy and helping creators make a living working in the metaverse,” he says. “We expect to be meaningfully better at monetization than others in this space.”

For the uninitiated: Horizon is what Meta is calling its first baby steps into the metaverse. Quest is the Meta-made virtual reality headset that now carries the Meta branding. And the Quest home—since renamed Horizon Home—is a VR living room where you and your friends, represented by your digital avatars, may someday hang out, selecting games or other virtual worlds to explore.

But this metaverse of several years from now is not what analysts on the earnings call want to talk about. When the operator opens the phone lines for Q&A, the first two questions are about the company’s struggles with short-form video, and how much the company needs to spend on artificial intelligence. Then there are questions about the company’s dramatically slowing revenue and earnings growth. The analysts want to talk about flatlining user numbers in the U.S. and Europe. They want to talk about how changes to both Apple’s iOS and Google’s Android operating systems—which those companies have touted as privacy enhancements—will make it harder for Meta to deliver value to advertisers. They want to talk about looming government antitrust and privacy actions on both sides of the Atlantic that could pose an existential threat to the company. 

Although Zuckerberg has spoken about Meta’s business challenges in his introductory remarks, when it comes time for these analysts’ pointed questions, he passes most of the queries to Dave Wehner, the company’s chief financial officer, and Sheryl Sandberg, its chief operating officer (who has since announced she’s stepping down). What the duo says boils down to: Yes, there are near-term challenges, but the company is optimistic it can overcome them. Zuckerberg, for his part, chimes in on how Meta will soon have an A.I.-driven recommendation engine for video to compete with TikTok’s (although he did not mention the rival social media company by name). The CEO also says that the gap between Meta’s initial investment to build its metaverse products and the company seeing real revenue from it “is going to be long”—longer, in fact, than for any product the company has built before. Analysts should take heart, Zuckerberg said, that the company was “laying the groundwork for what I expect to be a very exciting 2030s.”

But for Meta investors, the 2030s seem a very long way off indeed. The company’s stock price has dropped almost 40%—erasing some $349 billion in market value—since Facebook announced its rebranding as Meta in late October. (That’s more than triple the percentage decline the broader market has experienced over the same period.) That cratering stock price makes investors plenty nervous. It also disquiets Meta’s employees, who are partly compensated with stock—and makes them more likely to jump ship.

To shore up its earnings in the near-term, especially as the prospect of a recession looms, Meta has reportedly been forced to impose a hiring freeze in some divisions of the company. It has also, according to Reuters, slowed the pace of hiring in its Reality Labs division—the very unit tasked with building the metaverse. (Meta notes that it is continuing to make high-priority hires—such as more machine learning engineers—and that the hiring slowdown comes after a period of torrid growth: It ended the first quarter of 2022 with 28% more full-time employees than it had a year earlier.)

This in a nutshell is Meta’s dilemma: Zuckerberg is trying to drive his company at breakneck speed into the metaverse, but his company’s tires may well run out of tread, if not rupture completely, before it gets there. Even more challenging is that the tech to fully deliver the metaverse doesn’t exist yet. There’s no clear consensus about whether the metaverse will actually be the next big thing in our digital lives, or, if it is, about whether Meta will be able to make the kind of money there that it has from its existing social media platforms. In other words, Meta doesn’t just have to repair its aging tires, it needs to build a whole new powertrain, all without losing speed. Martin Garner, the chief operating officer at market research firm CCS Insight and a close observer of the social media companies, says Meta is trying to pull off “a balancing act it has never had to do in the past.”

Conversations with dozens of senior executives who have recently left the company, several current high-level executives, and a survey of more than 1,000 employees still at Meta, paint a picture of a company in flux. As Fortune wrote last week, there’s a general sense of unease about what the metaverse pivot means for individual employees, along with significant concerns about the competition the company is facing from TikTok, as well as the impact of the privacy enhancements from Apple and Google are having and will have on the business.

Most employees remain confident in Zuckerberg’s leadership and tech instincts, and in the ability of the company to deliver the metaverse vision the founder and CEO has laid out. But doubts are starting to creep in. The interviews also suggest that burnout has become a major factor at Meta—fueled, in part, by the continual public criticism that Facebook has absorbed in the wake of its repeated, high-profile missteps. A picture emerges of a company that has yet to fully reckon with those missteps, and risks repeating them—only this time in virtual reality.

A pivotal day

When the history of Meta is written, October 28, 2021, the day the company announced its rebrand from Facebook, will certainly get a major mention. Another turning-point date will be September 13, 2021, the day The Wall Street Journal published the first in a series of damning articles based on the leak of thousands of internal documents, collectively branded “The Facebook Files.” The series portrayed a company whose leadership was well-aware that its platforms were contributing to problems ranging from political misinformation to teen depression—and yet failed to take effective action to address them, often because doing so would hurt revenues or curtail growth. (Meta disputed the Journal’s characterization of the leaked documents.)

But for Meta as a business, a more momentous date is April 26, 2021. That is the day Apple implemented “app tracking transparency,” a feature that prevents companies such as Meta from monitoring iPhone users’ activity or tracking their behavior across apps. Users can still opt-in to tracking if they wish, but the anti-tracking feature is turned on by default.

While Apple has couched these changes as privacy enhancements for its users, Meta says that’s disingenuous, because Apple itself is still allowed to collect vast troves of individual user data—giving itself a potential edge over rivals in personalizing services and offering targeted advertising. “Time and time again, Apple makes changes that benefit their bottom line while harming consumer choice, businesses, and developers,” a Meta spokesperson told Fortune. Meta also took aim at Apple in a filing to the National Telecommunications Information Administration, part of the U.S. Department of Commerce, in response to that agency’s request for public comment on competition in the mobile app ecosystem. Apple’s new rules, Meta said in its submission, had helped boost Apple’s own ads business at the expense of others and also driven some app developers to shift from a business model based on ad revenue to one based on charging users a subscription, on which Apple then takes a 15% to 30% commission. (NTIA is working on a report that could help inform antitrust action by the Biden Administration.)

App tracking transparency (or ATT as it is known) is, for the most part, not a technological fix that prevents data from being transmitted to apps. Instead, ATT is a policy change that affects what companies like Meta are allowed to do with that data once they have it. ATT says that app developers must get permission to use this information to track an individual across across third-party sites and websites and serve them targeted ads. Apps, such as those owned by Meta, are still allowed, however, to receive most of the same information about a user’s behavior. It is just that they are now required to aggregate that data into larger buckets and only analyze or track the behavior of this larger group of people.

These changes at Apple nonetheless pose a significant challenge to Meta’s revenues, because they have made it much more difficult to insert relevant ads at the right time into users’ social feeds. The changes have also made it much harder to determine whether someone who saw an advertisement on one of Meta’s platforms subsequently purchased that product through a different app or website.

Patrick T. Fallon—AFP/Getty Images

While the same change is also affecting other social media companies that rely on advertising revenue—“the Tim Cook recession,” one former Meta executive dubbed it—Meta has taken a worse hit than most. That’s because Meta had been so good at tracking its users across the Internet before. The company long has had a reputation for having the best ad-delivery system in the industry, based in part on a tracking widget it invented, a small piece of software code called the Pixel. Advertisers can place the Pixel on their own websites, and it enables them to tell whether someone landed there after having previously seen an ad on one of Meta’s platforms and gather a lot of data about what the person does on that third-party website or app. It could also help that same advertiser target relevant users on Meta’s social feeds. Without access to this kind of data from Pixel and other tools, Meta can’t offer advertisers the same assurances that their campaigns are effective and can’t be sure it is capturing conversions accurately.

“They have no f**king idea who is browsing anymore,” says Tyler Horner, director of integrated media at marketing and branding firm Within. “And Meta’s whole power—the reason that their algorithm had such an edge—was that they were always analyzing people’s browsing data to figure out what they were in market for.”

The hit to Meta’s topline is substantial: Wehner, Meta’s CFO, estimated in February that Apple’s privacy changes will cost the company about $10 billion in lost revenue in 2022. That revelation, along with the company missing analysts’ quarterly revenue targets, contributed to a 20% drop in Meta’s stock price the very next day. In April, the company again missed analysts’ revenue forecasts, for the first quarter of 2022, and it also guided analysts toward lower revenue estimates for the second quarter. Its first quarter revenue growth, at 6.6% year-over-year, was the slowest it had ever reported in a decade as a public company.

To make matters worse, this February Google said it would follow Apple’s lead and restrict ad tracking on its Android operating system by default, beginning in 2024. (Android, which is the operating system used by more than 70% of the world’s smartphones, already introduced an opt-out from some ad tracking last year, but it is not enabled by default.) One former Meta marketing executive estimates that the combined impact of Apple’s and Android’s changes, as well as Europe’s tough data privacy laws, could leave the company facing a 30% decline in the effectiveness of advertising across its platforms.

Fighting ‘signal loss’ for advertisers

Nicola Mendelsohn is one of the leaders tasked with combating this decline. Mendelsohn is vice president of Meta’s global business group, which is responsible for the company’s relationship with advertisers; she was formerly the company’s top ad exec in Europe. She tells Fortune her teams are talking to customers about ways to overcome the “signal loss” Apple’s privacy changes caused. “We are working with them to help them navigate the new landscape,” she says.

Meta has tried several approaches to counter the Apple hit. One is to increasingly encourage advertisers to use a feature that integrates Messenger, WhatsApp, and Instagram direct messaging services with the ads it shows on Instagram and Facebook—enabling a user to click on an ad and open an immediate conversation over WhatsApp or Messenger with one of the advertisers’ sales reps, or at least with an automated sales chat bot. This provides advertisers a direct link between a user seeing an ad and embarking down the funnel towards a purchase— ultimately both increasing conversions and making those conversions simple to track. 

But for most advertisers, Meta increasingly relies on statistical models of the advertiser’s conversions, based on historical data, the behavior of aggregated pools of users with similar traits, and the data of users who have opted in to allowing Meta to track them. Sheryl Sandberg, the soon-to-be-departed executive widely credited with building the Facebook ad juggernaut, told investors on the company’s earnings call in April that such modeling can help advertisers “understand measurement and campaign performance, even when we’re not able to see or aggregate certain conversions.” The company is also trying to perfect A.I. software that could learn to effectively target ads and track conversions using much less data than current methods require. And it is researching new methods for using anonymized data and aggregated data that can improve measurement of ad campaigns while also preserving individual privacy.

Many digital platforms use modeling to track ad performance. A lot of advertisers, however, don’t trust Meta’s current model-driven estimates. Horner says the fundamental assumption of the models—that the behavior of a user who opts out of being individually targeted is broadly similar to that of a user with similar attributes, such as device type, browser type, geography and demographics, who has opted in—is faulty. He says there are likely essential behavioral differences between anyone who elects not to be covered by Apple’s ATT and everyone else who is covered that the model cannot capture. He also criticizes Meta for not being transparent with advertisers about how its models work. And Horner says that some of the ways Meta currently counts conversions seem bogus, as the company includes anyone exposed to a Meta ad in their feed who later makes a purchase as a conversion, even if there is no evidence the user paused long enough to view the ad, as opposed to having just scrolled past.

Meta says that its method for counting impressions without regard to how long a user dwells on an ad is common in the digital advertising industry. It says that advertisers can elect to count conversions based only on actual click-throughs if they wish, and that they can specify customized time windows—such as 24 hours or seven days—in which a purchase must take place for a conversion to be counted. Meta also says that it calibrates its models by comparing users who opt in to data sharing to aggregate data from the users who are opted-out of data sharing by ATT.

Another approach Meta is using is to try to convince advertisers to hand over their own marketing data, such as customer email addresses, IP addresses, and information from customer databases that might log in-store purchases too, so that Meta can then use that data to retroactively attribute conversions to ad campaigns. Customers can allow Meta to access this data through Meta’s Conversions application protocol interface (API). Meta declined to provide a figure to Fortune on what percentage of its advertisers are currently using the Conversions API or the related Conversions gateway. Meta says that in the months following Apple’s privacy changes, it was underreporting conversions by about 15% but that it has since brought this underreporting down to around 8%.

The Conversions API is not a panacea. Not all advertisers want to share their data with Meta, for starters. There also may be issues around complying with data privacy laws, such as Europe’s stringent General Data Protection Regulation (GDPR), if advertisers share customer data. To avoid violating the law, advertisers need to get user consent to share this data with Meta.

Beyond the Conversions API, Meta’s Mendelsohn says, “We’ve seen significant improvements on our measurement and our campaign performances as well, even when we’re not able to aggregate certain conversions.” She says that Meta’s platforms—Facebook, Instagram, and Whatsapp— still trump any rivals’ for enabling companies to find new customers. “No other platform generates demand for products and interest in brands like we do, and that is because of the 3.6 billion people that are using our services every day,” she says.

In conjunction with other companies, such as Mozilla, Meta has been researching ways of tracking conversions that would preserve user privacy, and the company has begun rolling out some of these methods in limited pilots. It has been testing a tool for called Private Conversion Lift that helps advertisers measure the incremental impact of an ad campaign on conversions without revealing individual user identities. It is also working with the World Federation of Advertisers on the creation of a standardized way to measure advertising impact across different media, including digital channels and television. The proposal includes the use of panels of people whose activity would be monitored—similar to the way the Nielsen audience measurement metrics have worked for television—combined with impression data from the digital platforms.

Because of the amount of data that Meta has already gathered, including third-party data that comes from advertisers or other entities, GDPR could present a huge regulatory risk to the company, according to a half-dozen former senior Meta executives Fortune spoke with. All requested anonymity to avoid violating non-disclosure agreements or alienating such a powerful Silicon Valley company. For the past two decades, “Data was flying everywhere, and more and more data was being tracked,” one former Meta engineering executive says. He says that Meta’s systems were simply not built for a world in which data usage must be carefully controlled based on the geography of the user and the preferences they have selected, and in which a lot of personal identifiable information can’t be used at all.

Motherboard, the tech news site that is part of Vice Media, recently published an internal Meta memo written by a privacy engineer assigned to its ads business, in which the engineer bemoaned the company’s inability to govern the vast amount of user data it has. The engineering executive Fortune spoke to says the leaked memo is significant and could open the company up to regulatory and legal action, including massive fines under GDPR—the law allows privacy regulators to fine violating companies up to 4% of global revenue. 

In a statement to Fortune, a Meta said the Motherboard story and the former engineer’s comments “misrepresent how we manage data.” It said it has “built one of the most comprehensive privacy programs to oversee data use across all of our operations” and that the document Motherboard obtained “was never intended to capture all of the processes we have in place to comply with privacy regulations around the world or to fully represent how our data practices and controls work. With regulations and privacy expectations evolving, we constantly assess risks and explore ways to meet our obligations more efficiently, which is what this document shows.”

Another former Meta executive who worked on regulatory issues told Fortune that inside Meta, there’s generally a belief that the company’s huge size and vast resources will enable it to “more nimbly comply” with new privacy regulations than smaller competitors. This perception is also evident among current Meta employees: in a survey of more than 1,000 current Meta employees conducted by the anonymous polling app Blind exclusively for Fortune, 81% said they were confident Meta could successfully comply with different privacy laws being developed worldwide.

Chasing TikTok

There’s another area where Meta, despite its size and resources, is clearly playing catch-up: short-form video. Here the company is seen as trailing far behind rival TikTok, which is owned by Chinese Internet giant ByteDance, both in terms of content and its appeal among the young people who are an especially attractive target for advertisers. To be sure, Meta’s short-form video feature, which it calls Reels and which launched in 2020, is rapidly gaining popularity among existing users, now accounting for 20% of the time they spend on Instagram, according to the company. But a recent survey from equity research firm Piper Sandler found TikTok was the social media app preferred by a third of all American teenagers, just ahead of Snap, at 31%, while Meta’s Instagram ranked a distant third, preferred by just 22%.

Meta is also struggling to convince advertisers to embrace Reels, and to persuade them of its advantages compared to its original News Feed and Stories text and image-based formats. Garner, the CCS Insight analyst, says part of the problem is that there is simply “less room for advertising” on short-form video than in Meta’s News Feed. Brian Nowak, an analyst at Morgan Stanley, has estimated that the monetization of Reels could still be 45% lower—meaning that Reels will deliver 45% fewer ad impressions per user—than that of Facebook and Instagram by the end of 2023.

TikTok hasn’t exactly figured this out either: The platform made only $3.88 billion in advertising revenue in 2021, or barely one-thirtieth of what Meta generated overall that year. But TikTok was expected to more than triple its figure this year to $11 billion, a number higher than the combined advertising revenue of Twitter and Snap combined, according to research firm Insider Intelligence. 

And TikTok’s user growth is explosive: The app is the world’s most downloaded so far this year, with more than 176 million downloads through mid-April. By the end of the first quarter of 2022, it had 1.6 billion monthly users, according to app tracking firm data.ai, a gain of more than 50% since September 2021, as its growth blew through previous forecasts. Meta, meanwhile, has been dealing with slowing user growth: In February the company reported for the first time in its history that it had experienced a quarter-over-quarter decline in daily active users of Facebook in Q4 2021, dropping slightly to 1.929 billion, compared to 1.930 billion. The daily active number recovered to 1.96 billion in the first quarter of 2022, although the number of users in Europe declined—something the company blamed on the war in Ukraine and the blocking of Facebook in Russia.

Step Up—Getty Images

Mendelsohn says Meta is working hard to help advertisers do more with Reels. The company is keen to monetize the video feature for several reasons. One is that its users are spending more of their time on it compared to other parts of Meta’s platforms. But it is also true that having advertisers run ads in multiple formats across Meta’s platforms is one way the company has suggested companies can recoup some of the data lost, and conversions missed, due to Apple ATT changes. Mendelsohn says the company is experimenting with different ad formats—overlays, banner ads, and stickers. And she says her teams regularly conduct “Reels School” where they put a mobile phone in the hand of a chief marketing officer or chief executive officer and get them to make Reels in order to show them “how easy it is to do it, how brilliant the suite of tools is, for both businesses and consumers to be able to play with it together.” 

Mendelsohn also says Meta is working with professional “creators” to help them make money from Reels, with the idea being that incentivizing these professional short-form video makers to do more on the platform will increase the amount of content available, draw in more users, and then also attract more advertisers.

Trusting the algorithm

Mark Zuckerberg told investors in April that Meta is preparing to change its recommendation algorithm for short-form videos to help surface viral content even if it is not being shared by people in a users’ immediate social network. This will make Meta’s short-form video recommendation system much more like the one credited with TikTok’s explosive growth. But it also seemingly marks a philosophical departure from the decision Meta made in early 2018 when Zuckerberg said he was changing the algorithm that controlled users’ Facebook News Feed. At the time, the CEO said he was “changing the goal I give our product teams from focusing on helping you find relevant content to helping you have more meaningful social interactions.” 

Now it’s back to relevant content, and not just when it comes to video. On the April earnings call, Zuckerberg said Meta is building an A.I.-powered “Discovery Engine [that] can show you all of the most interesting content that people have shared across our systems,” including Facebook posts and Instagram photos. “In the future, I think that people will increasingly turn to A.I.-based Discovery Engines to entertain them, teach them things, and connect them with people who share their interests,” he said. 

That language puts a Zuckerbergian techno-utopian polish on a change Meta is making out of pure competitive necessity. But it is also worth remembering that having Meta’s algorithm designed to find engaging or viral content, even if it wasn’t being shared by friends, was one of the things that critics charged made Meta culpable for political polarization, misinformation, and extremism, especially in the run up to the 2016 U.S. presidential election. That kind of content tends to be highly “viral” and “engaging,” which is why the company had moved away from this system in 2018. Put another way, Meta may be harkening back to its Facebook past to secure its short-term future, even as the mistrust sown by that past continues to be a drag on the company’s public image.

Meta is eager to send the message that the algorithmic changes do not mean the company is headed back to the Wild West days before 2018. The company says that while algorithmic modifications will surface more content that users might be interested in from outside their immediate social network, those changes don’t affect the recommendation system’s existing, post-2018 emphasis on content being shared by a user’s own social connections. Nor, Meta says, does the change remove or erode guardrails the company has put in place to prevent harmful, low-quality, or false and misleading content from being promoted to users.

Still, the focus on engagement in deciding what to recommend to users has lead Meta into all kinds of problems before. And issues such as online safety, misinformation, hate speech, and the promotion of extremism—may be poised to haunt it again as it pivots to the metaverse.

In our next segment, we’ll examine the technical challenges of building that virtual reality world, and how Meta hopes to enable its metaverse to escape the kind of backlash that has dogged its existing platforms.