A load test run at 11:00 on a quiet Tuesday rarely mirrors UK evening traffic, payday spikes, or Sunday retail windows. Shape scenarios with the calendar your product actually lives on.
Include authentication bursts, cache cold starts after deploys, and the batch jobs that collide with human traffic. Capture not only p95 latency but queue depth and dependency error rates that foreshadow collapse.
Treat the test report as a conversation artefact for product and ops, not a CI badge. If leadership cannot map the scenario to a real launch date, the test will not change capacity decisions.
