Performance Metrics interview questions — Hard
250 hard-level Performance Metrics questions from our Performance Testing bank, each with the correct answer and an explanation of why it is correct.
What this covers
Performance Metrics appears throughout Performance Testing interviews. At hard level, interviewers are typically checking depth under pressure — edge cases, failure modes, performance characteristics, and the trade-offs you accepted. Work through these, then explain your answer out loud; the second part is what interviews actually test.
4 example questions
1. Regarding Metrics, which of the following is a false statement?
- A. Baseline measurements enable comparing performance across releases.
- B. Standard deviation and percentiles describe response-time variability, not just central tendency.
- C. Averages mislead because a few very slow requests can hide behind a good mean.
- D. Tail latency is irrelevant for systems with many users.correct
Why: The selected statement is false. The other options are accurate statements about Metrics.
2. Identify the INCORRECT statement about Metrics:
- A. P95 latency means 95% of requests failed.correct
- B. P99 (tail latency) matters because at scale 1% of requests is many users.
- C. Error rate is the percentage of failed requests during a test.
- D. Averages mislead because a few very slow requests can hide behind a good mean.
Why: The selected statement is false. The other options are accurate statements about Metrics.
3. Which of the following statements about Metrics is FALSE?
- A. P99 (tail latency) matters because at scale 1% of requests is many users.
- B. Throughput and latency are the same metric with different names.correct
- C. Apdex scores summarize user satisfaction from response-time thresholds.
- D. The p95 latency is the value below which 95% of requests complete.
Why: The selected statement is false. The other options are accurate statements about Metrics.
4. Which statement about Metrics is NOT correct?
- A. Apdex scores summarize user satisfaction from response-time thresholds.
- B. Baseline measurements enable comparing performance across releases.
- C. The average is always the most reliable single latency statistic.correct
- D. P99 (tail latency) matters because at scale 1% of requests is many users.
Why: The selected statement is false. The other options are accurate statements about Metrics.