Describe leading vs lagging indicators with technical performance examples.

WHAT THIS TESTS: This question tests whether you understand metric taxonomy well enough to drive a feature with data rather than just shipping code. Interviewers want to see that you can tell the difference between a signal you can act on today and a result you can only observe tomorrow. Senior engineers are expected to instrument systems with leading indicators so the team can course-correct before users feel pain.

A GOOD ANSWER COVERS: First, a crisp definition of causality and time. A leading indicator is a measurable input or early-system signal that predicts a future outcome, while a lagging indicator is the outcome itself, measured after the work is done. Second, concrete technical examples tied to a performance feature. For leading, pick something like cache hit rate, connection pool saturation, queue depth, or garbage collection pressure. For lagging, pick something like P99 latency, error rate, throughput, or user-reported performance tickets. Third, the explicit linkage between them. You should explain how raising the cache hit rate is expected to drive down P99 latency, and why you watch the leading metric in a daily standup but review the lagging metric in a quarterly review. Fourth, a note on actionability. Leading indicators are useful because you can assign an owner and a target, whereas lagging indicators are useful because they prove business value.

COMMON WRONG ANSWERS: Confusing activity with impact is the biggest red flag. Saying deploy frequency or lines of code is a leading indicator shows you do not understand the concept. Another red flag is giving only business or financial metrics such as revenue or churn without technical translation. Some candidates also describe lagging indicators as bad and leading indicators as good; the interviewer wants to see that both are necessary and serve different decision-making time horizons. A weaker answer will list two metrics without explaining the causal relationship between them.

LIKELY FOLLOW-UPS: The interviewer may ask how you would set targets for a leading indicator when the lagging result is noisy. They might ask what you would do if the leading indicator improved but the lagging indicator did not, which tests your understanding of false positives and hidden variables. Another common follow-up is how you would balance short-term leading metric gains against long-term lagging metric trade-offs, such as aggressive caching improving latency but increasing stale-read errors.

ONE CONCRETE EXAMPLE: Imagine you are building a predictive prefetching feature to improve mobile app load times. A leading indicator would be the prefetch hit ratio, measured as the percentage of user navigations that find data already in local cache. You can A/B test this in days and tune the prediction algorithm. The lagging indicator would be the cold-start P99 latency measured from real user sessions over the following month. If prefetch hit ratio rises from forty percent to seventy percent but P99 latency stays flat, you know the cached assets were not on the critical rendering path, so you refine the model rather than declaring victory.

Source: Wikipedia: Performance indicator