Mobile User Lifecycle State Machine

Lifecycle Levers

Push notification intensity 40%

Onboarding quality 55%

Simulation speed 2.0 days/s

Cohort State

Day

Retention (non-churned)

New

Active

Risk

Dorm

Churn

How it works

Each of the 220 installs is a user in one of five lifecycle states, and every simulated day it makes one probabilistic transition — a discrete-time Markov chain, the same kind of model growth teams use to classify users by recency/frequency and target win-back campaigns:

New  --p_onboard-->  Active
New  --(1-p_onboard)-->  Churned (absorbing)

Active  --0.12·(1-0.5·push)-->  At-Risk
Active  --0.03 + fatigue-->      Churned
Active  --stays-->                Active

At-Risk --0.25 + 0.5·push-->  Active   (win-back via notification)
At-Risk --0.35·(1-0.4·push)-->  Dormant
At-Risk --0.08 + fatigue-->     Churned

Dormant --0.05 + 0.25·push-->  Active   (resurrection)
Dormant --0.10-->                Churned
Dormant --stays-->                Dormant

fatigue = max(0, push-0.6) · 0.4   (over-messaging opt-outs)

Push notification intensity — raises the daily probability that an At-Risk or Dormant user is pulled back to Active, but past ~60% intensity it also adds a fatigue term that pushes Active/At-Risk users straight to Churned (uninstall from notification spam).
Onboarding quality — sets the day-0 chance a brand-new install ever reaches Active at all; the biggest single lever on long-run retention.
Churned is absorbing — once a sphere turns red it never receives another notification or transitions again, exactly like an uninstall.
Every marker drifts to the hub of its current state each day; the five hubs are arranged as a pentagon so you can watch the whole cohort's population redistribute in real time instead of reading a single decaying curve.

New Active At-Risk Dormant Churned

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.