⚠ Off-Platform Signal LossModerate threat

Meta's behavioral signal is richest inside its own apps, but a real part of its historic advantage came from following people around the wider internet — the pixels and trackers on other websites and apps that told Meta what its users did after they left. Privacy changes and regulation have been cutting that off-platform signal away, and with it a slice of the picture Meta once assembled of each person's full digital life.

The danger is that a narrower view means less precise prediction at exactly the moments that matter most to advertisers — the off-platform purchase, the competitor's website visited, the intent expressed elsewhere. If Meta can see only what happens inside its walls, its model of the customer is powerful but incomplete, and incompleteness costs targeting accuracy and therefore price.

The consolation is just how much happens inside those walls. Billions of people spending hours a day on Meta's apps generate enough first-party signal that the loss of off-platform data, while real, is a trimming rather than a gutting — and AI has proven able to extract more predictive power from the in-app data than older methods could, partly compensating for what was lost outside.

That leaves a moderate risk. Off-platform signal loss is a genuine and permanent narrowing of Meta's view of its users, most costly for tracking intent that occurs elsewhere — but the first-party ocean is deep enough — Meta rebuilt on first-party and modeled signals after ATT¹ — and AI good enough at mining it, that the core prediction advantage endures.