At a glance: Use Prediction to project your app's or game's LTV using historical data. LTV Prediction is crucial to understand growth and forecast revenue for overall profitability.
To predict LTV, you will need to specify the target day from the initial open for prediction This will be referred to as the "Target day". The term "Predicted Target LTV" will be used to denote this forecasted value.
1. Go to Your App > Prediction
2. In the primary filters, select the "Date range" for your prediction data
3. Define the "Target day" for predicted LTV
4. Optional: You can choose to filter prediction based on Campaign and/or Country
Note:
A minimum of 5 days of complete data with at least 200 users in the filter dimension is required for accurate prediction calculation.
Example:
Data retrieved after filtering by "Campaign A" and "United States (US)" should have at least 200 users.
5. Information on Attributed eCPI and Attributed Installs are displayed for reference.
Ad Network / Attributed eCPI - Displayed for ROAS calculation using Attributed eCPI
Ad Network / Attributed Installs - Allows you to check if there is enough data for prediction
6. Click on "Apply" to generate the charts below.
7. Expand on each LTV Prediction chart to analyse Predicted Target LTV.
Using Monetisation Curve
Monetization curve analyses historical data to determine how much each day's LTV contributes as a percentage to the LTV of target day.
Example:
If the target day is D30 with an LTV of $1, then each day's LTV leading up to it is calculated as a percentage of this total. For instance, if the LTV on D0 is $0.2, it represents 20% of the total LTV on D30.
By leveraging historical data, Cost Center calculates the minimum, maximum, and average percentages of LTV D0 relative to the LTV of the target day. These percentages are used to predict the target LTV.
How to use:
Image: Monetisation Curve - Secondary Filters
To compare different data for Monetisation curve with its LTV chart below, you can define secondary filters for: App, Date Range, and Country.
This updates the Monetisation curve based on the additional filters while allowing you to make cross-comparison with the LTV graph generated from the primary filters.
Toggle to the "Detail" view to display breakdown by day.
Or "Split by Campaign" to view breakdown by campaigns.
Note:
Data points with fewer than 100 installs per day are excluded to minimize interference from outliers (min attributed installs per day).
Using Formula (Retention Rate X ARPDAU)
ARPDAU (Average Revenue Per Daily Active User) is a metric that measures the average revenue generated by each active user daily.
Here's an explanation of the LTV calculation based on Retention Rate and ARPDAU:
Initial Calculation (D0): Let's assume 100 users first open an app on D0. If the ARPDAU on D0 is $2, then the LTV on D0 is also $2 since each user generated $2 in revenue on this day.
Subsequent Day (D1): On D1, out of the initial 100 users, only 48 remain active. If ARPDAU on D1 is $1, it implies that these 48 users have contributed $1 * 48 = $48 in revenue on that day.
Increment in LTV (D1): The increment in LTV on D1 is calculated as the average revenue generated by all users installed on D0 (not by active users on D1). Therefore, it is $48 divided by the total number of users installed, which is 100, resulting in $0.48.
Calculation of LTV (D1): Thus, the LTV on D1 becomes the LTV on D0 plus this $0.48 increment, resulting in $2.48.
Generalized Formula for calculating LTV on any given day (D(x)) is as follows:
LTVD(x) = LTVD(x−1) + Retention Rate D(x)×ARPDAU D(x)
This formula considers the previous day's LTV, the retention rate on the current day, and the ARPDAU on the current day to calculate the LTV for the current day.
Using Power function
Power function LTV Prediction uses trend analysis that draws a trend line through historical data points to identify patterns or trends. This trend line reflects the overarching trajectory of LTV over time across various customer cohorts, enabling predictions about future LTV values.