SKAN Conversion Values

SKAN Conversion Values are the small piece of post-install information you are allowed to encode inside SKAdNetwork to learn what a user did after they installed, within Apple's privacy limits and timers.

You want to know more than "this campaign got an install." Did that user finish onboarding, hit a paywall, buy something? SKAN lets you answer a sliver of that through a conversion value, and you get two flavors. There is a fine value, a single number from 0 to 63, and a coarse value that is just low, medium or high, which is what Apple falls back to when your campaign is too small to protect anonymity with the fine number.

Because that is all you get, you have to design a conversion value schema up front. You decide which post-install actions map to which numbers, so that when Apple's postback arrives days later, those few bits tell you the most useful thing possible about who those installers really were.

Respect how brutal the constraints are. The bit budget is tiny and the measurement timers are short and unforgiving, so you cannot track everything. You are forced to choose which signals matter most, and a schema that wastes its bits on trivial events leaves you blind to what your highest-value users, the ones who might spend real rupees on day three, actually did. Treat the schema as a real design task, map it to revenue and retention, and revisit it when your funnel changes.

Spend your handful of bits on the actions that predict a paying user, and nothing else.