Mobile Attribution guide - Part 2 :-Mobile Attribution methods

Below are some of the methods currently used. There is no method which is the best, I will explain later in this article why I say so.

Before we proceed with the attribution, lets see how the attribution works , below is an example:

Example scenario- Click attribution window- 7 days

Here it takes 5 different channels(some repeated) over time for a user to finally convert. If the original display banner was not exposed to the user, he could have spent $1000 with any other competitor. every channel assisted in conversion someway or the other. Attribution gives a holistic view of the assisted and non assisted conversions.

For App install, we can see which network or partner actually helped to drive an install.

Conversion path

Display>Social>Search>Search>Direct

Available  Attribution models in Google Analytics- Last click, Last non-direct, Last adwords click, first interaction, linear, time decay and position based

Universal App Attribution Methods

Method 1: Google Install referrer method

Google Install referrer method

Attribution flow

1.User clicks on the ad run by one of the Ad network - &referrer= parameter with a unique click ID (ym_tracking_id) is appended to the link 

2.User is redirected to the attribution provider where referral and unique identifier is generated

3.It then moves to the Google play store , referral and unique identifier is also passed on

4.App downloaded> google adds referrer and identifier to the app

5.User open the app for the first time,SDK data and the database is matched

6.Attribution   is done as per the click data

Note:

(i) 1:1 accuracy -10% loss (ii) Supports customizable attribution window (iii)only click based attribution support (iv)Compatible only with Google play only, cannot track Android out of store

Method 2: Unique Identifier method

Attribution flow

1.User clicks on the ad run by one of the Ad network

2.When a user clicks the ad, the advertising network inserts various parameters in the tracking URL according to a template. One of these parameters is the device ID.Click is passed on to the tracker tool and is recorded with the device id bound to it

3.User is redirected to the play store

4.Downloads the app

5.When the user opens the app for the 1st time,, the tracking tool’s SDK collects the device identifier and send it to the app tracking tool platform with device identifier . Simultaneously the platform also receives an install request

6.The platform then searches for the clicks and installs that matches the description

7.Attributes the install to the partner based on this unique identifier. Any configurable postbacks are distributed

Note:

(i) 1:1 accuracy -10% loss (ii) Supports customizable attribution window (iii)Both click and impression  based attribution support (iv)Compatible only iOS and Android (v)Most common identifiers- Apple(IFA) and Google (AID) (vi)Not supported by all networks

Method 3- Device Fingerprinting method

Attribution flow

1.User clicks on the ad run by one of the Ad network

2.Click is passed on along with the data points to the tracker tool and is recorded. It collects the publicly available information from HTTP headers. Analytics platform creates a unique device fingerprint based in each click

3.User is redirected to the play store

4.Downloads the app

5.When the user opens the app for the 1st time,  the tracking tool’s SDK collects the same  data points and pass it to the analytics platform

6.The platform then searches for the clicks and installs with non-unique information and matches it with the

7.Attributes the install to the partner based on this unique identifier.

8.Any configurable postbacks are distributed

Notes:

(i) Least reliable method (ii)Data points- Platform, Device brand, Device model, Device carrier, IP Address, OS name, OS version, User agent, timestamp (iv) Non-changeable 24 hours attribution window(for some tools)(v)last click attribution

Method 4- Open URL with Click id method

Attribution flow

1.User clicks on the ad run by one of the Ad network. Tracker tool is notified about the click alng with the click id

2.As the app is previously installed, they go directly to the page on the banner- through deep linking

3.When the app is open, the SDK collects the open URL parameters including click_id

4.It then sends this to the App platform

5.Attribution Analytics platform takes the Click ID, finds the corresponding click record, and attributes the event to the click accordingly.

Notes:

(i)  Does not work for app installs. Only work if the app is previously installed (ii) Used for event tracking (iii) 1:1 accuracy (iv) It works for both app-to-app and web-to-app scenarios (v) offers customizable attribution window (vi) zero loss