A note on Mobile App Attribution Method

In an ideal scenario, the traffic source would pass a Device ID as a unique identifier for an individual device. At every click unique click id is generated, which is similar to gclid in google ads and fbclid for facebook ads. The combination of the unique click id and a unique device id results in an accurate attribution method.

In real world, mobile ads can be served either on third party apps or on mobile web browsers or in-app browser.

Scenario 1: When ad is served on third party apps

•  Post click on ads, device id is combined with MMP click id and stored with MMP to match it with installs or in-app events

Scenario 2: When ad is served on Mobile Web Browser or In-App Browser

•  Unique click id is generated
•  Device ID is not generated (device id is only generated for In-App traffic sources). Therefore, device cannot be identified
•  In-the absence of device id – MMP uses fingerprint and probabilistic technique to identify each device
• MMP combines IP address with device attributes like device name, device type, network provider, location, OS, OS version etc and behaviour attributes like CTIT, session duration etc of anonymous user to identify a device on ad click
• MMP uses this fingerprint information and probabilistic attribution algorithm to match click to install and attributes install to click source