What is Adjust?

What Adjust does

Adjust performs the standard MMP role. Its SDK sits in the app and reports installs and in-app events; advertising networks send their click and impression data to Adjust; Adjust matches the two under one consistent rule set and produces a de-duplicated view of which sources drove which outcomes.

Alongside attribution it provides cohort analytics, fraud detection, deep linking and handling for Apple's privacy frameworks — SKAdNetwork and AdAttributionKit postback processing among them.

Why an app would use one

The central argument is neutrality. An app buying across many networks otherwise receives many self-reported numbers from parties paid on the result, which sum to more installs than actually occurred — see biased attribution.

A single measurement layer removes the double counting and gives one comparable view. It also concentrates fraud detection: patterns like click spam and click injection are far easier to identify when one system observes every source at once.

Where the model is constrained

Every MMP shares the same limits, and they are properties of the measurement problem rather than of any vendor. Large self-attributing networks do not share raw interaction data, so their contribution is reconciled rather than independently verified. Unclicked influence and cross-device journeys remain invisible. And on iOS much of the output is aggregated privacy-framework data rather than deterministic matching.

None of that argues against using one. It argues for pairing it with a causal instrument — see incrementality — rather than treating any attribution product as ground truth.