Cohort Analysis

Ran cohort analysis on workout app usage by users based on signup date and referral source, with the goal of answering the following questions:

• How would you assess the retention patterns of this app? Are they good or concerning? Is there room for improvement?
• What are some seasonal or ‘over time’ patterns you’re seeing in the data?
• How would you define the ‘habit’ moment of this app?
• What are the key differences between acquisition sources when it comes to retention?

Findings

• There was an increase in signups during the weeks of 11/21 and 11/28 was mostly driven by paid signups, which is a source with the lowest retention on average anyway. These two cohorts did show the worst retention, even among that source. So, our campaign got a lot of signups, but 60-65% had dropped off by W7. (Note that this was also around the Thanksgiving holiday, which may play a role here.)
• Overall, referral signups have the best retention in terms of the percentage of members retained by W7. However, that is also our lowest source by volume, so it's a small proportion of signups.
• Referral signups seem to show a different retention curve that doesn't level off as much as the others but continues decreasing. So, referral signups may be a leaky bucket to look into. The good news is that they are sticking around in the platform for a longer period of time, so there is more chance of them having that 'a-ha moment'.
• The 10/10 cohort had better retention than others on average for both organic and paid sources. So, it may be worth looking into whether anything was unique about that week and that cohort.
• The habit moment, where users are likely to stay, is around week 3. That is where the curve starts to level off.