Using the techniques in How to Project Customer Retention by Fader & Hardie (2006), and an implementation of those techniques by JD Maturen, Retentionizer will fit a shifted-beta-geometric distribution to the data, show the projected retention rates for each cohort, show the imputed beta distribution for each cohort, and calculate the LTV of a given customer in that cohort.
Basically, it turns a sample of cohort survival rates:
| t | past 30 |
|---|---|
| 0 | 1.0 |
| 1 | .81 |
| 2 | .80 |
| 3 | .76 |
| 4 | .75 |
| 5 | .72 |
| 6 | .70 |
| 7 | .67 |
| 8 | .66 |
| 9 | .65 |
| 10 | .64 |
Into this:
Retentionizer is built by David Chudzicki and Chris Clark.