A/B Testing
a simple way to test changes
current design VS new design
enables data-backed decisions
shift business conversations from ...
“we think” to ...
“we know.”
Data-Driven Decision
... no opinion-driven, ego-driven, or authority-driven
determine the optimum number of results to display on a search engine results page
in ~2000
http://en.wikipedia.org/wiki/A/B_testing#History
A/B Test
= Split Test
= Bucket Test
Matrix
- clicks
- sign-ups
- downloads
- purchases / subscribes
Examples
A/B/C/D test
So, do we need A/B Testing?
Is our system today ready to support A/B Testing?
if not, what's missing?
A/B Testing Tools
Content Experiments
http://www.google.com/analytics/
A/B Testing on Azure
http://blogs.msdn.com/b/tomholl/archive/2014/11/10/a-b-testing-with-azure-websites.aspx