The first step in any experiment is to diagnose a measurable pain point. We used three sources of information to identify our main issues:
We monitored these three channels continuously to track progress and uncover any additional issues. While it may be tempting to address every issue reported by a user, having a measurable product metric (such as visits, click-through rate, add-to-cart, or purchases) helped us prioritize which problem to solve first.
Having established the baseline performance of any page or feature on our site allowed us to identify several solutions to address the opportunity and evaluate whether the expected impact was achieved through a new design. The beginning of each test started with a hypothesis statement. Here's an example.
Hypothesis statement
AB Testing is for learning. The featured cases allowed us to:
Capsules Product List Page — Original vs New We measured whether a more visual design and navigation increased add-to-bag and coffee discovery.
Machine Product List Page — Adding machine Subscription visibility We monitored the effect of adding subscription badges on the visibility of machines with available subscription plans.