All Collections
Frequently Asked Questions (FAQs)
Understanding data mapping in Cost Center
Understanding data mapping in Cost Center
Updated over a week ago

Cost Center processes user-level data to calculate Retention Rate, LTV, Play Time and other post-installation metrics. User-level data comes mainly from MMP, Firebase (via BigQuery) and Mediation services. Cost Center only gets Attribution data from MMP (i.e. which user belongs to which campaign). Analytics data including user activities and events are from Firebase while Ad Revenue is from the mediation.

Note

Cost Center does not get LTV, Retention Rate, ROAS, SKAN data directly from MMP.

Data matching algorithm: A user is identified by their ID: Advertising ID (GAID for Android, IDFA for iOS) or Vendor ID (same name for Android, IDFV for iOS). Cost Center uses this ID to match Attribution, Analytics and Mediation data. Cost Center's Retention Rate (RR) may not always match with the one from MMP because of the following reasons:

1. Difference in the number of attributed installs: While Cost Center gets Attribution data from MMP, there are several validation methodology to verify if a user is genuine:

a) Users without ID will be ignored.

b) Users who were active during the last 30 days will not be counted as new users. Cost Center uses Firebase data to check for active users.

2. Cost Center uses first-open events from Firebase to count new users, then matches it with Attribution data to calculate the number of Attributed installs for each campaign.

In some cases, Firebase’s first-open event may be wrong, this is checked against data from the last 30 days. If a user has their first-open event and wasn’t active during the last 30 days, they will be counted. Therefore, if a user was attributed to a campaign, then didn’t use the app for 30 days and Firebase fires a first-open event for this user, they will be counted again.

3. To calculate the Retention Rate, Cost Center uses data from Firebase, this may differ from analytics data tracking by MMP. The RR is calculated when a user is active in a day. For example, if there are 100 attributed users, 30 users have at least 1 event on Day 1, then RR at Day 1 is 30%.

4. Cost Center only calculates RR Day X when there is enough data for a user on Day X. For example, assume today is 10/8 and the data is imported until 9/8, then there is only data for RR Day 1 for users installed on 09/08 and before that.

Note

The number of installs in MMP always matches the number of installs in the Ad Networks (for non-SKAN ad networks like Unity, Mintegral, Applovin etc). This is because they are the same source. The ad networks use posts back from the MMP to count the number of installs.

Did this answer your question?