Your question is Debugging Performance Bottlenecks. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How do you approach debugging a performance bottleneck in a system where you do not have full visibility into the client's internal stack?
Explain how you would structure the investigation, collaborate with the client, gather actionable evidence, prioritize hypotheses, and protect delivery timelines. Include how you would communicate uncertainty, define escalation points, and validate that a proposed fix improves performance without introducing regressions.