How would you coach a team with fluctuating velocity?

This question tests if you use metrics for coaching, not just reporting. A great answer first reframes the goal from *high* to *predictable* velocity to enable forecasting. Then, you'd investigate qualitative data (retros, blockers, story quality) and quantitative data (work types, cycle time). Finally, you propose small, concrete experiments. A major red flag is suggesting ways to 'game' the numbers or using velocity as a performance metric to be maximized, rather than a diagnostic tool for the team.