Question for anyone who buys users for a mobile app.
When a new campaign looks bad in week one, how long do you give it before you cut it? And what do you judge it on: installs, trials, first purchases, or do you wait for renewals?
We ask because we used to cut on week-one numbers at our studio, and kept finding out months later that some of those weeks would have paid back.
When we replayed two years of our own spend, the weeks our model said to cut returned $0.55 per $1 by day 90, against $1.51 for the weeks it kept.

Curious what rules others use. We'll share what we've learned in the replies.