On 24/10/15 02:30, Alex Davis wrote: > Why do we need to ping every time the app opens and we don't just rely > on install numbers? We know for a fact that not all installs are equal. > If we only optimize for installs, we may ignore that cheaper installs > via 45-65 year olds actually have a really bad churn rate, and that > despite being more expensive per install, 18-25 years retain much > better. In the end, we don't want to optimize ads to drive installs, we > want to optimize our spending around acquiring more Firefox users.
Why do we need data this detailed in order to get a good enough idea of churn rates? What if e.g. we pinged on the first open every week; if someone didn't open the app that week, no ping. The distance through the week could give us a bit of a proxy for frequency of usage. I think we could get a good idea of which channels were better, without tracking every app open by the user. In general, I think we should always be thinking "what's the minimum amount of data I need to collect to get a good-enough answer to my question". Did we approach this from that perspective? Gerv _______________________________________________ mobile-firefox-dev mailing list [email protected] https://mail.mozilla.org/listinfo/mobile-firefox-dev
