Effective 3rd Party In-App Analytics Integrations

In-app analytics are what help your business understand how your apps are being used. All too often, integrating 3rd party analytics tools into your app is done hastily and without the kind of care that your team would take for a user-facing feature or major architectural concern. This can lead to both data issues and code maintainability challenges. Included here is a simple process for planning and implementing 3rd party in-app analytics tools for your apps.

Transcript

1. May 1, 2018
   Effective 3rd Party In-App
   Analytics Integrations
   

   View Slide

2. 2
   What do I mean by In-App Analytics Tools?
   ● Mixpanel
   ● Flurry
   ● Localytics
   ● Swrve
   ● Amplitude
   ● Google Analytics for Firebase
   ● Adobe Marketing Cloud
   Tools for measuring what users are doing in your app
   

   View Slide

3. 3
   Why Care about In-App Analytics Implementation?
   ● You don’t want to think about analytics all the time
   ● You will have to update/add to it over time
   ● Mistakes have consequences for your business
   

   View Slide

4. 4
   Why Care about In-App Analytics Implementation?
   ● Happy Developers
   ○ Cleaner app code
   ○ Easy to add/modify/replace tools
   ● Happy Business/Product People
   ○ More accurate information
   ○ Better product/business decisions
   

   View Slide

5. 5
   TEST
   Give yourself and your business partners
   confidence in the data you are providing.
   Overview
   PLAN
   Business/product and engineering need to
   be on the same page about what will be
   measured and how it will be measured.
   DOCUMENT THE PLAN AND
   IMPLEMENTATION
   The analytics implementation needs to be
   documented where both business and
   engineering can access it in the future.
   IMPLEMENT
   Make it easy on future developers (including
   yourself) by loosely coupling your app with
   your analytics tools and introducing
   constants for strings used in analytics
   measurement.
   2
   3
   4
   1
   

   View Slide

6. 6
   If your product/business partner didn’t tell you what,
   specifically, they need to measure and why, then STOP
   INTEGRATING and go ask them.
   1. Have a Plan
   IMPORTANT
   

   View Slide

7. 7
   1. Have a Plan
   Start by understanding what actions your users take that
   are most closely tied to success/revenue
   ● Move outward from there to significant steps before and
   after that
   ● Measure more than just your key actions, but don’t try
   to measure everything
   Product/business should talk with engineering about what
   is going to be hard to measure accurately--is it worth it?
   IF YOU DON’T KNOW WHERE TO START
   

   View Slide

8. 8
   1. Have a Plan
   ● List people/profile data that needs to be
   tracked/available
   ○ Examples: age, gender, order count, card member type, timezone,
   paid subscriber (true/false)
   ● List each event name, the desired metadata for those
   events, and when/where you plan to measure
   ○ Example: Measure each “social share” event with “social platform”
   and “content type” metadata AFTER callback for successful
   content share.
   ● Plan to be consistent with naming across platforms
   (web, iOS, Android)
   WHAT TO INCLUDE IN YOUR PLAN
   

   View Slide

9. 9
   2. Document the Plan
   Business people AND developers need to know what
   these event names are and what they mean.
   ● A spreadsheet works well or shared wiki
   ● Track *when* data started being collected
   ○ Example: Data available since version X.X.X, but “event foo” was
   first measured in version X.X.Y
   IT’S KIND OF LIKE A CONTRACT
   

   View Slide

10. View Slide

11. 11
    3. Implement
    MyAwesomeAnalyticsEvents {
    Const String EVENT_SOCIAL_SHARE = “Social Share”
    Const String EVENT_PURCHASE = “Purchase”
    }
    MyAwesomeAnalyticsProperties {
    Const String PROPERTY_SOCIAL_PLATFORM = “Social Network”
    Const String PROPERTY_SHARE_CONTENT_TYPE = “Shared Content Type”
    Const String PROPERTY_MEMBER_TYPE = “Shared Content Type”
    }
    CREATE CONSTANTS
    

    View Slide

12. 12
    As much as possible, avoid using 3rd party
    classes/methods directly in your app code.
    Instead, create a wrapper for all initialization, profile
    updating, and event measurement:
    3. Implement
    CREATE AN ANALYTICS WRAPPER
    

    View Slide

13. 13
    3. Implement
    MyAwesomeAnalyticsFacade.initializeAll()
    MyAwesomeAnalyticsFacade.setMemberType(“gold”)
    MyAwesomeAnalyticsFacade.incrementTotalOrders()
    MyAwesomeAnalyticsFacade.measureSocialShareEvent({PROPERTY_
    SOCIAL_PLATFORM : “twitter”, PROPERTY_SHARE_CONTENT_TYPE:
    “image”})
    CREATE AN ANALYTICS WRAPPER
    

    View Slide

14. 14
    3. Implement
    MyAwesomeAnalyticsFacade {
    initializeAll() {
    Mixpanel.init(MY_MIXPANEL_KEY, ...)
    Flurry.initialize(MY_FLURRY_KEY,...)
    }
    setMemberType(String value) {
    Mixpanel.setProperty(PROPERTY_MEMBER_TYPE, value)
    Flurry.setUserProperty(PROPERTY_MEMBER_TYPE, value)
    }
    measureSocialShareEvent(Map properties) {
    Mixpanel.measureEvent(EVENT_SOCIAL_SHARE, ...)
    Flurry.measureEvent(EVENT_SOCIAL_SHARE, ...)
    }
    }
    CREATE AN ANALYTICS WRAPPER
    

    View Slide

15. 15
    4. Test
    ● Yes, test just like you’d test any other code.
    ○ This is best done manually
    ○ Use your spreadsheet as a guide
    ● Have a debug/developer account with your
    analytics provider to keep your prod data clean
    TEST YOUR IMPLEMENTATION
    

    View Slide

16. 16
    TEST
    Give yourself and your business partners
    confidence in the data you are providing.
    Use your spreadsheet/plan as a testing
    guide.
    Summary
    PLAN
    Business/product and engineering need to
    be on the same page about what will be
    measured and how it will be measured.
    DOCUMENT THE PLAN AND
    IMPLEMENTATION
    The analytics implementation needs to be
    documented where both business and
    engineering can access/update it in the
    future.
    IMPLEMENT
    Make it easy on future developers (including
    yourself) by loosely coupling your app with
    your analytics tools (make a wrapper/facade)
    and introducing constants for strings used in
    analytics measurement.
    2
    3
    4
    1
    

    View Slide

17. View Slide