You've shipped a new product feature and need a clean way to evaluate whether it is working. The team wants one primary success metric, but also wants supporting metrics that capture early user response and protect against harmful side effects.
How would you choose the primary metric and guardrail metrics for a new feature launch?