This is the 4th post in a series of 6 posts. Here is a summary post explaining all the different calculators.
The method explained in the 3rd post assume you have 180 days of data which is sometime not possible. Here you can find a calculator that uses 90 days of data from existing segments to model out the 180 day LTV of the existing segment. It then leverages the same ratio method to calculate the life-time-value in the new segment.
- Training Data from existing segment:
- ARPU for existing segment users – day-1 to day-7
- ARPU for existing segment users – day-1 to day-90
- 7-day retention of existing segment
- 90-day retention of existing segment
- ARPDAU of existing segment – days 75-90
- Segment data:
- ARPU for new segment users – day-1 to day-7
- CLV presented as a number and in a Gauge
This model has 2 steps:
Step 1 – Estimating the LTV for 180 days – this is done by combining the known ARPU from the first 90 days with estimated ARPU for days 91 to 180. The estimation is based on multiplying the ARPDAU in day-90 by the expected user days from day-90 to day-180. The user-days calculation is based on a power function model.
Step 2 – Applying a ratio
Once we have the Estimated 180d LTV for the existing segments we use a simple ratio to estimate the LTV in the new segment:
- Divide the new segment 7d ARPU by the existing segments 7d ARPU
- Apply the same ratio to the 180d LTV in the existing segments
- The result is your 180d LTV in the new segment
- Allows calculation of LTV for newer apps
- Very accurate
- A bit complex to follow
Advanced hack – honning the model with new data
If you have more than 7 days of data from the new segment you can actually use any number of days as long as you are switching the input for both 7 day figures:
- Input the x-day ARPU in existing segments in the 7-day ARPU field for exsiting segments
- Input the x-day ARPU in new segment in the 7-day ARPU field for new segment
As long as ‘x’ is the same the model applies.
More methods to calculate user value
Here is a simpler method that can be applied if you have data from 180 days on existing segments
If you have time and a spreadsheet you can also fully model out the retention function