Constructing a hypothesis

A key step in developing and designing an experiment is constructing a hypothesis. To help you structure your hypothesis in a meaningful way, here are two helpful templates -- a simple one, and an advanced one.

Simple Hypothesis Template

1. Because we observed motivating data or feedback
2. We expect that change will cause expected metric impact

Advanced Hypothesis Template

1. Because we observed qualified and/or quantified data
2. We expect that change for cohort will cause an increase/decrease to metric as measured by avg|sum|count|% unique of event per traffic type
3. We expect to see metric delta(s) over a period of x business cycles