Create each menu change as an experiment with item, action, stores, channels, dates, owner, hypothesis, success metric, and stop threshold. Record whether a concurrent control group and a significance test exist.
Label before-and-after differences as observations, not causal effects.