The $10 Billion Meta Blamed on Apple Was a Number Nobody Could Verify

On February 2, 2022, a chief financial officer said a number on an earnings call, and within hours it had become the most repeated statistic in the history of digital advertising: Apple's privacy change would cost Meta $10 billion this year.¹ The market wiped roughly a quarter-trillion dollars off Meta's value the next day. The number traveled the world. It got attributed to Zuckerberg, hardened into fact, and turned into the morality tale of the decade — the data giant, finally taught a lesson by the privacy champion. There was just one problem. The man who said it, in the same breath, said he couldn't actually stand behind it.

The official story is that Apple's App Tracking Transparency cost Meta $10 billion. The truer story is that Meta's CFO offered a forward-looking guess about an unfinished year, called it a guess, and then watched a guess become gospel — because a single vivid number told a cleaner story than the messy truth underneath it.

The thing Apple actually broke

Start with the part that is real, because it is real. On April 26, 2021, Apple shipped iOS 14.5, and with it the enforcement of App Tracking Transparency — not the announcement a year earlier, but the live, mandatory prompt.⁴ The switch was deceptively small: iOS went from opt-out to opt-in for access to the device's advertising identifier. Before, an app could read the IDFA by default and you had to dig into settings to stop it. After, every app had to ask, in plain words, on a system screen, and most people, when asked plainly, say no.

That single flip cut the cable Meta's machine ran on. Meta's ad engine was never really about showing you ads on Facebook; it was about following you off Facebook — seeing that you tapped an ad, then bought the shoes three apps later, and feeding that loop back to advertisers as proof the ad worked. Strip away the identifier and the loop goes dark. You can still buy the ad; you just can't prove it sold the shoes. Attribution is the product, and Apple turned off attribution. The geography of the iPhone did the suppressing for them — Apple didn't have to argue with Meta, it just had to change a default screen on the most valuable customers in the world.

Why $10 billion was a story, not a measurement

Here is the detail almost everyone skipped. At the very moment Wehner forecast the $10 billion headwind, Meta's business was not contracting — it was setting records. Q4 2021 ad revenue came in at $32.6 billion, up 20% year over year, the quarter that crossed Meta past $100 billion in annual revenue for the first time.⁷ The $10 billion was not a wound you could point to in the financials. It was a projection of a future that hadn't happened yet — a counterfactual about how much more Meta would have made in 2022 in a world where Apple had left the IDFA alone. Counterfactuals don't show up in audits. They show up in narratives.

When the actual pain arrived in 2022, Meta's own paperwork told a far more crowded story than the headline did. The company's 2022 10-K — the document that is legally bound to be honest — names 'the more challenging macroeconomic environment and limitations on our ad targeting and measurement tools, among other factors.'² Read that clause slowly: macro first, ATT second, and 'among other factors' carrying the rest. A year later, on the Q4 2022 call, CFO Susan Li was even plainer. ATT was 'still certainly an absolute headwind,' she said — but the decline was driven primarily by 'weak advertising demand' from an 'uncertain and volatile macroeconomic landscape.'³ The single villain had quietly become one suspect among several.

Why would a company volunteer a scary number about itself? Because a tidy, external villain is the most useful thing a struggling business can own. 2022 was the year recession fear gutted ad budgets, TikTok ate younger attention, Reels were monetizing worse than the Feed they were replacing, and Reality Labs was burning roughly $10 billion a year on a metaverse the market had no patience for. Every one of those is a story about Meta's own choices and Meta's own vulnerabilities. ATT is a story about Apple. Given the menu, the single most quotable line a CFO could hand the world was the one that pointed the blame somewhere else — and named a round, memorable figure to anchor it. The genius of the $10 billion wasn't that it was precise. It was that it was portable.

The fair objection: maybe the damage was even worse

The honest counter to all of this is that the skepticism cuts both ways. ATT really did break the targeting loop, and the structural blow was severe — severe enough that an unverified estimate could plausibly have been too low, not too high. And the popular sense of the damage was, if anything, inflated by a misleading statistic: the famous '95% of U.S. users opt out' number came from early-adoption data and described people who had already locked down their privacy settings before the prompt even appeared. Global opt-in rates among users actually shown the prompt later stabilized around 46% — meaning roughly half of prompted users said yes, far from the near-total blackout the '95%' implied.⁸ So the real targeting loss was somewhere between 'catastrophic' and 'serious,' and the round $10 billion sat comfortably inside that range. Calling it a story doesn't make it false. It makes it unverifiable — which is a different and more uncomfortable thing.

And there is a twist that makes the $10 billion stranger still. A 2025 London employment tribunal filing alleges that a former Meta product manager claims the company used 'deterministic matching' to link data across platforms — partially routing around ATT itself — and separately exaggerated some ad-performance figures by up to 19%.⁶ The claims are unproven allegations in a filing, not findings. But if even partly true, they point at an awkward possibility: that Meta was quietly mitigating the very headwind it was loudly quantifying. A company can be genuinely hurt by a rule and still have an incentive to overstate the hurt while privately blunting it. Both can be true at once. That is exactly why a self-reported, unaudited number deserves the asterisk it almost never gets.

Apple did real damage to Meta — that part needs no spin. It turned off the measurement that made Meta's machine worth more than anyone else's, and the structural cost will compound for years. But the $10 billion was never the wound. It was the dressing Meta chose to put over a much messier set of injuries, most of them self-inflicted. The lesson isn't that Apple won or that Meta lied. It's that the most expensive number in the story was the one no one could check — and that a great crisis response isn't the one that fixes the problem. It's the one that decides who gets blamed for it.