Why AdWords shows X conversions for a campaign and Analytics shows Y?
Why Facebook shows X conversions for a campaign and Analytics shows Y?
I have heard these type of questions more times than I have “Good Morning”
It’s a nightmare for analysts when two tracking systems report different metrics for same dimension and the nightmare gets even scarier with app analytics. Because often behavior analytics tools like google analytics are not enough to track app install sources and mobile measurement platforms like Apsalar, Apsflyer, Tune, etc. are used to track AppInstalls.
Why advertisers use mobile measurement platforms in addition to google analytics?
Postbacks: Ability of mobile measurement platforms to send Appinstall success to the publisher which sent the user to play store. This is especially more important because often publishers are paid cost per install rather than on traditional CPM / CPC models. Google analytics has also added postback functionality recently but still is limited to selected publishers.
Why advertisers use behavior analytics platforms in addition to mobile measurement platforms?
While mobile measurement platforms provide great insights at user acquisition, these often are found wanting when it comes to deep diving in-session behavior of user where traditional analytics tools like google analytics do a better job.
Nightmare starts when questions are asked about Install numbers according to MMP and “New Users” numbers in google analytics are not even in the same ballpark. One seems to be in football park and other in baseball park.
Lets take a look how users are defined for MMPs and for Google Analytics.
MMP: Most of MMPs will track install by AID which is semi-persistent device identifier (GAID/IDFA) and reinstall will not be counted as install.
Google Analytics: Definition of “New users” in Google Analytics has nothing to device identifier of any kind and is based on clientID which is generated by GA SDK. There is very little documentation available about when this ClientID changes for a device, so I conducted my own experiment and found out following.
- Launch the Application>>Kill the Application>> Launch again : Client ID does not change which means this user will not be treated as new user. (But it will change time between 2 App launches exceeds 2 years)
- Launch the Application >> Kill the Application >> Clear data for your app >> Launch again: Cliend ID for changes on 2nd launch which means this user will be considered a new user and acquisition source for this user will be direct
- Launch the Application >> Kill the Application >> Uninstall the application >> Launch again: Cliend ID for changes on 2nd launch which means this user will be considered a new user and acquisition source for this user will be direct.
This shows how Google Analytics may count same device as new user while MMP will not and GA “New users” will be more than installs number reported by MMP.
- User ID tracking: If your application requires users to log in you can use User ID tracking feature of GA to minimize this discrepancy
- If your app does not require users to log in then you can pass Advertising ID (GAID/IDFA) as User ID of device.
Starting career with paid search, Nikhil has built and managed award winning digital campaigns, has been Google Adwords Top Contributor, Planned and implemented several complex and large scale web and mobile analytics implementations. Currently working for Myntra as Manager Analytics Nikhil is also conducts trainings on Web Analytics, Mobile Analytics and Digital Marketing.