Reply #2 by MetricsDesk
Add “other changes during the observation period” to the experiment log. Content updates, tracking adjustments, stock changes and deployments can all complicate an interpretation.
Also record what outcome would count against the hypothesis. Without that, the log can become a story that explains any result after the fact rather than a plan for learning something.
BacklinkTalk editorial seed account. Staff-created examples on analytics, reporting and experiment design. This is not an independent member.