⚠ Attribution ErosionLow threat

Measurable results depend on attribution — the ability to trace a sale back to the ad that caused it — and the same privacy changes that blunt targeting also blur that trace. When Meta can no longer follow a user from an ad to a purchase on another company's website or app, its proof of a return becomes an estimate, and an estimate is less persuasive than a certainty. The closed loop that makes advertising feel like a science springs a leak.

The danger is compounded by skepticism about self-reported numbers. Meta grades its own homework — it measures and reports the results of ads sold on its own platform — and as attribution weakens and marketers grow more sophisticated, some question whether the reported returns are as good as claimed. A measurement moat that rests partly on trust is vulnerable if that trust erodes.

The offset: Meta has rebuilt much of its measurement on modeled, privacy-safe methods and AI-driven attribution that no longer need to track individuals, and its sheer scale gives it more data to model with than anyone. For the vast performance-advertiser base, the returns remain good enough, and demonstrably so, that budgets keep flowing.

File it under low-to-moderate. Attribution erosion genuinely makes Meta's proof of its own effectiveness fuzzier and invites more advertiser skepticism than in the tracking heyday — but modeled measurement and scale have largely filled the gap, and the results remain convincing enough to keep the budgets coming — the recovery after the ATT shock proved it¹.