Analytics
User clustering
Creating dashboards
Custom attribution modeling
Forecasting & prediction models
Analyzing advertising campaigns and improving them
CASE
ROAS coef prediction for a mobile game
Predictions help make decisions about changes in advertising campaigns or products based on data
More effectively allocate the budget
Challenge
calculating the ROAS coefficient in terms of platforms and types of monetization (ad revenue, in-app, and total) to adjust the KPIs
Issue
The old coefficients did not accurately reflect the real income picture, thus affecting the ROAS KPIs. With the old coefficients, for target ROAS values of 52% on day 7, breakeven (100% return) was achieved on day 90 of the cohort.
Our analysis resulted in new coefficients. Consequently, with these new coefficients, the day 7 ROAS should be at least 36% to achieve ROAS 100% on the 90th day.
This finding facilitates quicker achievement of one of the project’s key goals –– cost-effective scaling.
Methodology
1st
Took data from Appsflyer and divided the sample into two different monetization models, as we assume they will have different retention
2nd
Collected data on the total revenue and number of purchases, in order to calculate the average ticket size and retention
3rd
Looked at the shape of the retention curve by revenue count for the two different monetization models and for different OS
4th
Calculated the average ticket size for each of them.
5th
Calculated the LTV for a user + expected LTV with an “extended” curve
6th
Derived the coefficients per user (and per cohort) — dividing LTV from day 90-540 by LTV on the 7th day.
It turns out that the parameters for the model are the day 0 retention rate and day 0 average ticket size. If the average ticket size increases significantly, it is worth changing it for the entire cohort so that the predicted LTV is based on updated data.
The smarter way to achieve a higher ROAS
Profit-driven User Acquisition by seasoned marketing experts for your App and Website
Contact Us