Cohort Lab / 01
100 synthetic users

What makes users stay?

Tune four product forces, then watch a tiny cohort live through twelve weeks. No forecast theater—just a playful model you can poke.

Try this: run “Leaky bucket,” then raise weekly usefulness without touching acquisition. Watch what changes—and what doesn’t.
Starting condition
How many people struggle before finding the first useful moment.
The chance the product earns another visit this week.
Delighted users can pull a few waiting users into activity.
Higher pressure amplifies churn when value is still uncertain.
Week 0 / 12
Ready to simulate
0Active
0Delighted
0Churned
—Retained
New
Active
Delighted
Churned
Tap a dot—or focus and use arrow keys—to inspect

Retention, week by week

this run50% baseline
Choose a preset or tune the controls. Every run replays the same 100 synthetic users, so changes are comparable.

A deliberately simplified model. Useful for intuition, not prediction.