The live graph was the overview. After the run (or while it finishes), dig into correctness under concurrency — your pm.test() blocks still ran on every iteration.

1. On the finished run, confirm the summary still matches what you saw live (avg, error %, throughput — and p95 if available).
2. Open the per-request breakdown if shown. Compare Get Hot Coffee vs Get Iced Coffee — one endpoint often degrades first on a real API.
3. Check failed test assertions and errors. Status-code tests should stay green here; on a real system this is where racey or empty responses show up.
4. If you see Tests vs VU, skim it — it shows when assertions start failing as concurrency rises. Treat it as a preview, not homework.

Expected outcome: You can separate “the chart looked fine” from “the tests stayed green,” and name which request you’d watch first next time.

More: performance metrics · test assertions under load.