Retention metrics can be broken into two components – Frequency and core behaviour.
The distinction exists because we need to first identify what is it that someone does (action) that acts a metric for retention, then we also need to consider what core behaviour acts as a trigger that ‘locks-in’ the frequency.
Frequency need to be calibrated to the natural patterns of your product offering, daily login to Facebook is expected, daily shopping for insurance isn’t – so we need to ensure that frequency is determined based on what is reasonable for the kind of product.
It makes sense therefore that in instance where frequency is not fully determined you run historic data to see if patterns emerge, then use this as a baseline off of which to optimise.
Defining the core behaviour can be challenging, but its generally the magic moment – a point which demonstrates that users ‘get’ your value proposition, examples –
- Uber: Booking a trip
- Facebook: Viewing the news feed
- Slack: Messaging someone
From here you can map data on a line graph showing <time frequency> retention by core behaviour, with each line denoting a different core behaviour across a horizontal axis of time. The vertical Y axis is the % of the cohort that completes the action.
Further questions can then be explored such as – should you require doing the action X times? How do you define your frequency? Does weekly mean 3 of 4 weeks or 4 consecutive weeks?