Simulate a rule before enabling
Some automation actions are irreversible — sending emails, closing cases, firing webhooks, charging refunds. Simulate Execution lets you evaluate a rule against real data and see exactly which conversations it would match, without any of the actions actually running.
This is the safety net you should reach for before turning on any rule that does something to a caller, not just internal bookkeeping.
Before you start
- A rule built and saved (it doesn’t need to be enabled yet)
- Admin permissions to edit automation rules
Steps
-
Open the rule. Go to Settings → Automation Rules, find the rule, and open it in the editor.
-
Click Simulate Execution. The button is in the rule’s toolbar, near the Save button.
-
Pick a scope. Choose what data to evaluate against. Options usually include “all conversations from the last N days” or “conversations matching this filter” — pick the smallest scope that reflects what the rule will see in production.
-
Run the simulation. Atender evaluates the trigger and conditions against every conversation in scope. No actions run. No messages are sent, no statuses change, no webhooks fire.
-
Review the matches. The result lists every conversation that would have triggered the rule, plus which branch matched for each. Click into a conversation to confirm the match makes sense.
-
Adjust if needed. If too many or too few conversations matched, return to the conditions panel and tighten or loosen them. Re-run.
Verify it worked
The simulation result should show a sensible number of matches — neither zero (the rule’s too narrow or wired wrong) nor your entire inbox (the rule’s too broad). Spot-check several matched conversations to confirm they’re the kind of thing the rule should react to.
When to simulate
Always, before enabling, when the rule:
- Sends messages to callers (email templates, SMS, Slack DMs to customers)
- Closes or reopens cases
- Fires webhooks to external systems
- Reassigns ownership in a way that affects SLAs
You can skip simulation when the rule only adds tags, adds notes, or sends internal Slack pings — the cost of being wrong is low.
Simulate vs. Manual run
These two are different:
- Simulate Execution — evaluates the rule, lists matches, runs nothing. Safe.
- Manual run — evaluates the rule and executes the actions against the matches. Used for one-off backfills (e.g., “tag every existing returns conversation”). Walks you through Simulate first, then Execute.
If you mean to actually run the rule against historical conversations, see Run a rule on historical conversations.
Troubleshooting
-
Symptom: Zero matches. Fix: The conditions are too narrow, the trigger is wrong, or the scope window is too short. Widen the scope window first to rule that out.
-
Symptom: Far too many matches. Fix: Tighten conditions. The most common offender is forgetting
message.direction = inbound— without it, a content-matching rule fires on agent replies too.