> ## Documentation Index > Fetch the complete documentation index at: https://evalgate.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # Simulate, then activate a monitor > Prove the monitor on history before it touches live traffic. # Simulate, then activate a monitor A monitor goes through three states: **draft → simulated → active**. You never skip straight to active. ## Step 1 — Simulate On the monitor version page, click **Simulate**. EvalGate runs the scorer against historical traces that match the filter and shows you the would-be alert rate, score distribution, and any samples that would have fired. A screenshot of the simulation results showing the would-be alert rate. ## Step 2 — Adjust if needed If the simulation fires too often or never, adjust the threshold or scorer, save a new version, and simulate again. A monitor that never fires is as useless as one that always fires. ## Step 3 — Activate When the simulation looks right, click **Activate**. The monitor now samples live traffic, scores it, and can fire alerts. A screenshot of the Activate button on a simulated monitor. ## Step 4 — Pause or resume You can **pause** an active monitor at any time (for example, during an incident) and **resume** it later. Pausing keeps the version; it only stops sampling. ## Backfill If you need to score older traffic under the current monitor, use **Backfill** with a bounded time window. Backfill is explicit and bounded — it does not silently re-score everything. ## Up next When the monitor fires, [read the alert](./read-alert).