Choose the Success Metric for a Feature

What interviewers evaluate

• Do you tie the metric to the feature's goal and the company mission?
• Do you pick ONE primary metric, not a dashboard of vanity metrics?
• Do you include guardrail metrics to catch unintended harm?
• Do you reason about leading vs lagging indicators?
• Do you consider cannibalization and second-order effects?

A framework to structure your answer

1. Goal - why does this feature exist? What user/business outcome is it meant to drive?
2. Primary metric - choose one metric that best captures that outcome (an action tied to value, not a raw count).
3. Guardrails - name metrics that must NOT degrade (e.g. overall engagement, revenue, latency, the cannibalized surface).
4. Leading vs lagging - identify a leading indicator you can read quickly and the lagging outcome it should drive.
5. Cannibalization - check whether the feature just shifts behavior from elsewhere rather than adding value.

Try structuring your own answer first, then reveal a strong worked example.

Common variants

• What's the north-star metric for [WhatsApp / YouTube / LinkedIn]?
• How would you measure the health of a marketplace?
• Pick one metric to optimize the whole company around - which and why?

Pitfalls to avoid

• Choosing a vanity metric (raw clicks, page views) instead of one tied to value.
• Listing 10 metrics with no single primary.
• Forgetting guardrails - so a 'win' could be hiding harm.
• Ignoring cannibalization (the feature just moved engagement around).
• Picking a metric you can't actually measure or attribute.

Likely follow-ups

• Stories usage is high but overall time-spent is flat. Success or failure?
• How would you set up an experiment to attribute the lift?
• What guardrail would make you roll it back?