Engineering User Engagement

How does one build a product with over 10 Million users, at a sub Rs 1/- campaign cost per install, 90%+ organic users, with over 40% retention as well as 5 minutes per day of time spent? The answer for Mubble lay in the org design and OKRs that were set around some rather simple math, but was fiendishly difficult to execute. The difficulty comes from the inter-linkages and inter-dependencies across the various variables, and the resulting challenges in alignment as well as prioritization.
What I shall try to do is use a few posts to outline how we thought about the problem of user engagement, and how we designed our team structure, data collection & analysis, goals, tasks and review mechanisms around metrics that implied & allowed strategic coherence.
Engagement = Acquisitions * Retention * Active Users * Sessions Per Day * Time Per Session
This rather obvious equation helped us view engagement as a funnel. You acquire users, end up retaining some (who keep the app on the phone), of whom a few open the app, and they do so regularly, and spend time on the app each time they open the app. As I mentioned earlier, the grunt-work really is in selecting the right metrics to track, designing org structures around these metrics & creating an execution rhythm around the above.
For the purpose of this note, I have used a time scale of a single day. However one of our apps had a time-scale of a week, while another had a day. I guess it’s important to keep a perspective of time-scales that reflect the intersection of product intent & current/expected user behavior. But I’m getting ahead of myself…
Our structural design therefore morphed to teams that were dedicated to the following
- Acquisitions : Ad campaigns, Contests, Product Features (Virality Drivers), Network Effect Design
- Retention : Product Onboarding & Feature Adoption Cohorts
- Active Users : Habit Driven Usage (organic) & Notification Driven Usage (inorganic)
- Sessions Per Day : Frequency Cohorts
- Time Per Session : UX Design
Note that the above only led to engagement. We had a slightly different approach to monetization, which I shall try and explain in a subsequent post.
