Performance Trend
The Trend tab on a load test result shows how the same test has performed across its recent runs, so you can spot a creeping regression or confirm an improvement at a glance. It plots three charts over your run history: P95 response time, error rate and throughput.
Open the Trend tab
Open any result for a test, then click the Trend tab. The page builds itself from the test's run history (matched by test name), so the more often you run a test, the richer the trend. The Trend tab works the same way for cloud, JMeter and k6 tests.
Read the charts
Three charts are plotted, one point per run, oldest to newest:
- P95 response time (ms): the 95th-percentile response time of each run.
- Error rate (%): the overall error rate of each run.
- Throughput (req/s): the requests-per-second each run sustained.
Hover a point to see the exact value and which run it belongs to.
Choose how many runs to show
Use the Runs to show control to set how many of the most recent runs are plotted (2–50), then click Load. The status next to it tells you how many runs were found and drawn. A brand-new test with a single run shows a single point; the trend becomes more useful as runs accumulate.
Threshold lines and pass/fail colours
When pass/fail thresholds are set on the test, the Trend charts add two things:
- Threshold lines: your Max P95, Max error rate and Min throughput limits are drawn as reference lines, so you can see how close each run is to the limit.
- Pass/fail colours: each run's point is coloured green when it met your thresholds and red when it breached them.
Without thresholds the charts still plot the raw metrics; the lines and pass/fail colours only appear once thresholds are enabled. See Pass/Fail Thresholds to set them.
Why it matters
A single result tells you how one run did. The Trend tab turns your run history into a story: response times drifting up after a release, an error rate that spikes under a new traffic pattern, or throughput recovering after a fix. Run the test on a schedule and the Trend tab becomes an early-warning view for performance regressions.