What is a conversion value?

Why it exists

A privacy-preserving postback cannot carry arbitrary event data — rich detail would allow re-identification. What it can carry is a small number, with an agreed meaning that both sides understand in advance.

That number is the conversion value. It is the entire post-install signal available through the framework, which is why the design of the encoding matters more than almost any other measurement decision on iOS.

Designing a schema

You are compressing everything you want to know about a user's first days into a handful of bits. Some approaches survive contact with reality better than others.

• Encode revenue in buckets where monetisation is the goal — the bucket boundaries matter more than the granularity.
• Encode funnel depth where the product has a clear activation sequence.
• Encode a single high-signal event where one action predicts long-term value well. Simplest, and often the most useful.
• Do not encode everything. A schema that tries to capture five dimensions in a few bits reports none of them reliably.
• Fix it and leave it. Changing the schema breaks comparability with every prior period.

The timer and the threshold

Two mechanisms constrain what you actually receive.

The measurement period is bounded — the value can only be updated within a window after install, so anything that happens later cannot be represented at all. If your product's value emerges in week three, the framework cannot see it, and you need a proxy that appears earlier.

And the value is subject to crowd anonymity thresholds. Low-volume campaigns receive a coarse conversion value or none, which is another reason to consolidate campaigns rather than fragment them.