Revenue is a poor metric - lag between actions and outcome is measured in weeks if not months. It's good to track over months, but does nothing in short term, so is not actionable.
I want to know what is the expected "time to stop using the app" at any given day. If this time is getting longer my app is getting better, so this gives me a metric that's better than revenue. However, this metric can only be accurately computed in retrospect, so I need to mine for correlations with something that's available sooner.
Something like "people who use the app on the first day 5 minutes are likely to drop out in two months, but people who use it on the first day for 20 minutes are likely to stay for 6 months".
Revenue is a poor metric - lag between actions and outcome is measured in weeks if not months. It's good to track over months, but does nothing in short term, so is not actionable.
I want to know what is the expected "time to stop using the app" at any given day. If this time is getting longer my app is getting better, so this gives me a metric that's better than revenue. However, this metric can only be accurately computed in retrospect, so I need to mine for correlations with something that's available sooner.
Something like "people who use the app on the first day 5 minutes are likely to drop out in two months, but people who use it on the first day for 20 minutes are likely to stay for 6 months".