Defend a metric choice under scrutiny

Tell me about a time you selected a non-obvious primary metric (for example, time-per-session rather than total time) and a senior stakeholder challenged you.

1. How did you get ready to defend the choice (counter-metrics, backtests, risk analysis)?
2. Which tradeoffs did you accept, and which guardrails or secondary metrics did you propose?
3. How did you agree to revisit the decision later (pre-registered thresholds, stop-loss, or sunset criteria), and what did you do when early evidence went against your choice?

Overview: This question assesses a data scientist's competence in metric selection, stakeholder communication, and experiment governance when justifying a non-obvious primary metric.