Compare an experiment’s primary and secondary outcomes
Keep a favorable side result from replacing the decision you set out to make.
Write the hierarchy before launch
Name the primary outcome, secondary diagnostics, and guardrails. If the decision is checkout completion, clicks on a help link can explain behavior but do not replace completed orders.
Read results in that order
Check delivery and instrumentation, then the primary result and uncertainty. Review secondary outcomes for plausible explanations. A favorable metric discovered after looking at many results is a hypothesis worth checking.
Turn surprises into follow-up work
If support clicks fall but orders do not improve, investigate whether the page answered questions or hid help. Run a follow-up if the distinction would change the rollout decision.
Try a related tool
Find the share of eligible visits that completed a defined action. Count at most one conversion per visit for this calculation.
