Retention by Property

Retention by property means breaking the retention curve by a user or event attribute — entry channel, device, city or plan — to see which group retains better.

Total retention is an average, and an average can hide two opposite realities; breaking retention by property is what opens the average up. If channel A's users retain at 40% and channel B's at 10%, a 25% average shows neither and misleads the acquisition decision. The practical value of this break is directing budget: a channel bringing longer-staying users is worth more per entry, even at lower volume. A common trap is breaking on a property that is itself an effect of retention, not a cause (say 'purchase count', where both being high and staying come from loyalty); a useful property should be fixed at the cohort's start, like entry channel. Another trap is over-slicing until each group is so small its percentage is noisy and invalid.