Glossary

What is a coarse conversion value?

A coarse conversion value is the low-granularity signal Apple's privacy frameworks return when a campaign has too little volume to justify fine-grained reporting — typically low, medium or high rather than a specific value.

Attribution and measurement

All 220 terms

Why the data gets coarsened

Apple's frameworks release detail in proportion to how many users it could describe. A precise conversion value attached to a campaign with a handful of installs could identify an individual, so precision is withheld until volume makes that impossible.

The mechanism is crowd anonymity: the more installs a campaign produces, the more the framework will tell you about them. Coarse values are the middle tier — you learn something, but only roughly.

What you can still do with it

  • Compare campaigns by distribution. The proportion of high versus low is a genuine signal even without exact values.
  • Track the distribution over time rather than reading any single period's mix.
  • Design the schema so the buckets are decision-relevant — if low, medium and high map to meaningfully different actions, coarse data is enough to act on.
  • Consolidate spend to push campaigns above the threshold where it matters most.

The structural incentive

Coarsening creates a pressure that shapes iOS media buying generally: fewer, larger campaigns report better than many small ones. That runs against the granular testing habits digital marketers built in the identifier era, and it is not a temporary condition to wait out.

The teams that adapted stopped treating campaign structure as a targeting decision and started treating it as a measurement decision — accepting less granular control in exchange for data that actually arrives.

Common questions

Can I force fine-grained values?

Only by having more volume in the campaign. The threshold is Apple's and is not configurable; consolidating spend is the only lever you control.

Is a coarse value useless?

No, if the schema was designed for it. A three-bucket signal that separates good installs from bad ones supports most budget decisions; a schema needing fine values to mean anything degrades to noise.

More in attribution and measurement

Deciding which touchpoint earned the outcome, and proving it.