Simulate a rule before enabling
Some steps cannot be undone: an email that was sent, a closed case, a webhook call. Test on recent conversations replays a rule on your real conversations and shows which ones it would have run on. No step runs, and nothing is sent or changed.
Use it before you turn on a rule that does something to a customer, not only internal bookkeeping.
Before you start
- A rule open on the rule page. It does not need to be saved or turned on — the test uses the rule as it is on screen.
- Admin permissions to edit automation rules
Steps
-
Open the rule. Go to Settings → Automation rules and open the rule, or build a new one.
-
Click Test on recent conversations. The button is in the header, next to the save button.
-
Read the result. Atender replays the rule on recent conversations from the last 7 days. The result says how many times the rule would have run, and how many conversations were skipped. Each example shows Would run or Skipped, with the reason.
-
Check the email. If the rule has a Send an email step, the result shows the email it would send, with the fields filled in from a recent conversation. A Send email template step is not shown.
-
Change the rule if necessary. If too many or too few conversations match, change the conditions, or ask the assistant to change them. Then test again.
How the replay works
- A rule on a message trigger, such as A customer sends a message, replays the latest message of that type on each conversation.
- A rule on another conversation trigger replays each conversation as it is now.
- If no conversation in the last 7 days had the event, the test says there is nothing to replay yet.
Rules the test cannot replay
The test says so, and shows no result, when:
- The rule runs on a timer, for example A conversation has no activity for a while, An SLA is about to be breached or An SLA is breached.
- The rule checks what changed in the event, for example a condition on New conversation status or Tag added. What changed is not kept on the conversation.
- The event does not happen on a conversation, for example a schedule, case or incident trigger.
For these rules, turn the rule on and check Run history after the first runs.
Verify it worked
The result shows a sensible number of matches: not zero (the rule is too narrow or wrong) and not every conversation (the rule is too broad). Read several examples to make sure they are the kind of conversation the rule should act on.
When to test
Always, before you turn a rule on, when the rule:
- Sends messages or emails to customers
- Closes or reopens cases
- Sends webhooks to external systems
- Changes the assignment in a way that affects SLAs
You can skip the test when the rule only adds tags, adds notes or sends internal notifications. The cost of a mistake is low.
Rules that run by hand
A rule with the trigger Run by hand has its own tools on the rules list: Simulate this rule, and a run on existing conversations. See Run a rule on historical conversations.
Troubleshooting
-
Symptom: It would not have run on any conversation. Fix: The conditions are too narrow or the trigger is wrong. If every skipped conversation failed the same condition, the result says which one.
-
Symptom: Far too many matches. Fix: Add or tighten a condition. Check the trigger first: A customer sends a message never runs on a message an agent sent, so a rule that runs on agent replies has the wrong trigger.