Goal for this week (i.e. by Monday): implement a first stab at user cookieing, assignment to random “treatment”, and basic goal tracking, just in order to do A/A testing.

A/A testing (so called in the HiPPO article) is the practice of A/B testing where all the variants are actually the same. This may seem pointless, but this actually accomplishes a couple of goal: first, it establishes a baseline average for the goal metric (for my first quick sample, just a conversion rate). Second, it can confirm the validity of the cookieing and random assignment code. Finally, with enough data, it should confirm the null hypothesis, that is, that both treatments do equally well with respect to the goal.

Crucially, this means, for this week’s goal, I am not implementing the mechanism by which different treatments are produced.