Choosing the right mobile analytics solution can help you understand user behavior, engage users, and maximize user lifetime value. After this session, you will understand how you can learn more about your users and their behavior quickly across platforms with just one line of code using Amazon Mobile Analytics.
Scaling API-first – The story of a global engineering organization
(MBL303) Get Deeper Insights Using Amazon Mobile Analytics | AWS re:Invent 2014
1. November 12, 2014 | Las Vegas, NV
Andy Kelm, AWS Mobile
PatrikArnesson& VedadBabic, ForzaFootball
Chris Keyser, AWS Partner Program
2.
3. The King of England sent his best men ahead to learn context to plan the battle
4. Tactics used:
Segmented army into three divisions
Chose higher elevation around flat land
Waited for the enemy to buy time to rest and prepare
Built a system of ditches and pits to bring down the enemy cavalry
13. Amazon Cognito Amazon Mobile Analytics Amazon SNS Mobile Push
Kinesis DynamoDB S3 SQS SES
AWS Global Infrastructure (11 Regions, 51 Edge Locations)
Core Services
Mobile Optimized
Connectors
Mobile Optimized
Services
Your Mobile App, Game or Device App
AWS Mobile SDK
Compute Storage Networking Analytics Databases
Integrated SDK
22. 0
440,000
880,000
1,320,000
1,760,000
2,200,000
February 2012
October 2012
April 2013
November 2013
CHOOSE THE RIGHT TOOL
FROM THE BEGINNING
GA
(Google Analytics)
GA Mobile
GA sample data
24. THIS WAS IMPORTANT FOR US
WHEN CHOOSING ANALYTICS TOOL
Pricing
Flexible pricing
(Pay as you use)
Competitive pricing
Features
Retention
Custom events
Mobile friendly
Data
Ownership
Export functionality
No sampling
28. GROWTH PER COUNTRY
THE WORLD CUP IS THE MOST INTERESTING FOR THE AMERICANS
World Cup
29. PENETRATION
USERS PER CAPITA
Potential
Same penetration in the UK as in Denmark would equal 4 400 000 users
30. LEVERAGE ON MISSION
TABLEAU VISUALIZATION
The users in the nordic countries are the most interested in voting
31. HOW AND WHY DO USERS USE OUR APP?
DATA GUIDE DESIGN DECISIONS
32. PUSH NOTIFICATIONS
9/10 PEOPLE I HAVE MET SAY THAT THEY REMOVED THE APP BECAUSE OF TOO MANY NOTIFICATIONS.
33. NOTIFICATIONS
>30% set notifications for more than 11 teams.
We wanted to see if the amount of notifications could affect the retention.
0-4 notifications
5-9 notifications
10-14 notifications
37. SEGMENTATION BASED ON ACTIVITY
WE WANT TO SEE WHICH GROUP USE EACH FUTURE
Low activity
Medium activity
High activity
Build features for the medium activity group
38. APPROACH TO SEGMENTATION
WE WANT TO SEE WHICH GROUP USE EACH FUTURE
Experiment
Define criteria
User group size
40. CRITERIA
HOW TO DEFINE CRITERIA
Days active (14 days)
•# days active over 2-weekperiod
•Gaussian distribution?
•Let’s try it out!
41. USER GROUP SIZE
•Averaged over 16 time periods
•~60% in medium activity
•Result
•Low activity: 1-4 days
•Medium Activity: 5-12
•High activity: 13-14
HOW TO DEFINE SIZE