[Data/Analysis] How do you approach an interview question asking what metrics you would track after a process change?

Example Answer

I start with the objective of the change, because the right metrics depend on what the process is supposed to improve. If the goal is faster turnaround, cycle time and backlog may matter. If the goal is quality, error rate or rework may matter more. If the goal is better customer or internal-user experience, then completion rate, handoff success, or user effort might matter.

I also try to balance outcome metrics with operational metrics. Outcome metrics tell us whether the business problem improved. Operational metrics tell us whether the process is functioning in the way we expected. That matters because sometimes a change improves one number while quietly creating a different problem elsewhere.

If I am answering this in an interview without company context, I make my assumptions explicit. I would rather say, "Assuming the goal is reducing manual delays, I would look at these measures," than give a generic KPI list. The measures would remain provisional until the objective and operating context were confirmed.

I would define the unit, eligible population, numerator and denominator, timing boundaries and data owner before comparing periods. For example, 10 error cases among 100 eligible cases is 10%, while 8 among 40 is 20%: fewer errors by count but a higher error rate. I would compare like-for-like case mix and include guardrails such as rework or user effort. A before-and-after change alone does not establish that the process intervention caused the result.

Common Poor Answer to Avoid

"I would track efficiency, quality, and customer satisfaction."

Why it's weak

• It names broad metric categories without showing how the candidate connects measurement to the actual change being made.

Why this works

• It shows problem-based metric selection, tradeoff awareness, and comfort making assumptions explicit.