Tool 13 — Free, no sign-up
Fit the curve, then argue with it.
Three retention numbers is enough to fit a power curve, project D90 and D180, and turn ARPDAU into an LTV you can hold against your CPI. The fit is honest about how little data it had.
The link carries everything you typed. Nothing is sent to us.
Your numbers
Ads plus IAP, per daily active user. Blended across all regions.
The curve
Power fit a · d−b
Log fit (comparison)
Your observed points
| Day | Observed | Power fit | Log fit | Cum. active days | LTV |
|---|
LTV against CPI
Cumulative LTV
CPI
| Horizon | LTV | LTV / CPI | Verdict |
|---|
How it works / caveats
- The fit is an ordinary least-squares line through your points in log-log space: ln r = ln a − b · ln d. That gives r(d) = a · d^−b.
- b is how fast you leak players. Below 0.4 is sticky, above 0.6 is a leaky bucket.
- Cumulative active days to day N = 1 + the sum of r(d) for d = 1..N. Install day counts as one full day.
- LTV(N) = ARPDAU × cumulative active days. It assumes ARPDAU is flat across the life of the cohort, which is generous early and stingy late.
- The log fit is shown because people use it. It goes negative, which is why it is not driving anything here.
- Fitting three points and projecting to D180 is extrapolation, not measurement. The D90 number is a hypothesis with error bars you cannot see.
Retention is cohorted, not averaged. If your D7 came from dividing yesterday’s returning users by last week’s installs, throw it away and pull it properly.