Data Science, Product Analytics, Experimentations and more…

Are you switching back? The advanced version of A/B test you should know.

Switchback Experimentation Intro by HBS research

If you want to break into Marketplace industry as a data scientist, one of the areas that can help you optimize your experiments (usually A/B Test) is Switchback Experimentation.

Today’s article is an essence extracted from HSB’s publication of “Design and Analysis of Switchback Experiments” by Iavor Bojinov, David Simich-Levi and Jinglong Zhao.

Photo by Isabella and Zsa Fischer on Uplash

Why not do a typical A/B test?

As you may know, in multi-sided marketplace, there can be contaminations of experiments where typical A/B test cannot be done. Specifically those can be the 2 types of reasons:

1. Interference
2. Heterogeneous (or personalized effects)

Photo by Julia Koblitz on Uplash

So the challenges exist for A/B test, why not do switchback now?