Apple Tried To Destroy Meta’s Ad Business. Here’s How It Emerged Stronger Than Ever.

Leaning on artificial intelligence and other novel workarounds, Meta saved its ad business.

This post originally ran on Big Technology

In an instant, Apple brought the hammer down on Meta’s ad business.

After allowing developers to monitor people’s activity from app to app, the iPhone maker asked them if they wanted to be tracked. Most, unsurprisingly, said no.

That hit Meta right where it hurt. The social giant's secret weapon was its ability to target people with great precision — since most used their real identities on Facebook, Instagram and WhatsApp — and figure out whether the ads they saw led to purchases off Facebook. Without that tracking, the system basically broke.

After Apple’s move, Meta CEO Mark Zuckerberg said the company would take a $10 billion hit. To adjust, his engineers would basically need to rebuild its advertising technology from the ground up. A Wall Street Journal headline blared: “Inside Facebook’s $10 Billion Breakup With Advertisers” and featured a longtime Facebook advertiser telling a reporter, “it kind of feels like the end of an era.”

But then, somehow, Meta beat Apple’s restrictions and emerged stronger than before. Today, Meta’s stock price tops $700. And during its most recent earning call, it reported $40.59 billion in revenue, a 19% jump year over year, massive growth for a company if its size.

Instead of suffering, Meta has been roaring back since that initial $10 billion hit. A lot of companies would’ve been screwed. But the methods Meta employed, relying on artificial intelligence and other novel workarounds, might’ve saved its business.

Here’s how it pulled it off:

An Elegant Plan B

When Meta was faced with Apple's anti-tracking changes, conventional wisdom was that brands would no longer be able to tie what happens on other apps to ad views. When you see an ad for Ted's Homemade Socks on Facebook, and then buy those socks elsewhere, Meta would no longer be able to give that Facebook ad credit.

But Meta had a solid plan B. With the iOS changes, advertisers couldn’t track people across apps, but they could pass along their own conversion data (e.g. people visiting their sites, or making purchases) to Meta via a 'server-to-sever' connection, bypassing Apple entirely. Meta could then match up ad views with that conversion data and vice versa — making reasonable deductions about which ads drove which sales, visits, etc. This all took place within Meta’s conversions API — or CAPI — and delivered solid data about which ads were performing for advertisers.

"They realized for this to work, you need to understand the post-advertising experience," said JD Donahue, co-founder of the ad tech consultancy Up and to the Right.

But CAPI didn’t entirely solve Meta’s problem. Post Apple's shift, neither Meta, nor its advertisers, would be able to exactly zero in on specific consumers when putting together ad campaigns. So Meta began employing broader user "cohorts" for targeting and measuring on Apple devices.

In making this change, Meta went from being able to verify that a conversion occurred via their first party data to having to infer that a conversion occurred based on a model of consumer behavior. "What happened was they went from deterministic to probabilistic," said Joseph Yakuel, CEO of the ad agency Within.

"They lost some fidelity," added Donahue. And initially, ad performance suffered. "It went down, then started to build right up. [The new system] wasn't as good at first, then it got smarter."

There are some who contend that Meta's new suite of ad measurement tools aren't as effective as they used to be, but are still better than what's on the market. "Marketers needed it to work," said Sarah Livingston, a veteran ad tech consultant. "There was an entire ecosystem at stake. These brands rely on it. TikTok doesn't have the scale. Amazon works if you sell on Amazon…so If Facebook's performance was 15% less, it's still very good."

Analyst Andrew Lipsman, who authors the newsletter Media, Ads + Commerce, noted that Meta is only able to track ad conversions for seven days, versus 28 days in the old system. "In a lot of cases, you might see return on ad spend go from $4 to $8, and unsophisticated advertisers just accepted that number."

Still, Fred Leach, Vice President of Product Development at Meta, in a recent appearance on the MobileDevMemo podcast said that overall return on ad spend has improved 12% vs. 2022 — the year that everything supposedly went to hell.

Mobile consultant (and host of the MobileDevMemo podcast) Eric Seufert doesn't buy that average Meta advertisers are throwing around dumb money. "I categorically reject the idea that small businesses don't understand the performance of their ad campaigns — in fact, SMBs are very sensitive to the performance of their ad campaigns because ad spend is essentially a cost of revenue."

AI Comes Into the Picture

Meta also benefited from the rise of AI. About a year after Apple's ATT crackdown, Meta rolled out product called Advantage+. Similar to Google's Performance Max, Advantage+ is an AI-powered ad system via which brands supply Meta with goals, data and, other targeting information, and then let the machines take over media buying and optimization.

AI optimized buying is becoming an increasingly prominent part of the digital ad business, an area where only a few tech giants can play, and will likely maintain a massive advantage in a new era of AI-driven advertising. "You give it a goal, you feed it content, it finds the audiences which are most predictively going to achieve results," said Yakuel.

A great example is Kole Jax Designs, an ecommerce jewelry brand. The company has used Meta's lead gen ads and Advantage+, helping to drive sales up 70% year-over-year July 2023 to July 2024, and a ridiculous 665% yearly spike on Facebook and Instagram.

So AI plus CAPI seems to be a killer combo. Yet there are tradeoffs. "This has made ad products more opaque not less," said Yakuel. "In the case of Advantage+ it abstracts away the targeting options."

Brands therefore know less about how their campaigns work, and arguably have less control. Yet at the same time, Advantage+ works well. Really well, in part because the more brands use it, the more data they feed into the system.

"We've done dozens of A/B tests at WITHIN, comparing performance from a manually optimized conversion campaign vs Advantage+. And Advantage+ wins every time," said Yakuel. "It's a transition from audience-based advertising to outcome based advertising. Meta continues to be best in class at using its massive data advantage to drive tangible outcomes for brands."

So far, that has been working so well, Meta might even want to thank Apple.