Automations and campaigns

For

Managers and operators

Owner

Automation product

Outcome

Run repeatable customer work with visible safety and failure states.

Last verified

2026-08-30

Next review

2026-11-28

Status

supported

Customer journey

Automate safely

Estimated time

20 min

Permissions and roles

Company admin · Operations manager

Open the completion checklistSteps, proof, and recoveryClose checklist

Prerequisites

• A tested audience
• Clear ownership and stop conditions
• A healthy provider path and a human fallback owner

Permissions in practice

• Company admin: configure the automation, audience, and stop controls.
• Operations manager: approve ownership, test scope, human handoff, and rollback.

Steps

• Define the customer outcome, trigger, audience, owner, limits, and stop condition.
• Test the automation with synthetic records and inspect each transition.
• Confirm human handoff, provider readiness, failure evidence, and duplicate protection.
• Start with the smallest approved audience and monitor execution state.
• Pause, recover, or roll back when evidence does not match the expected result.

Success proof

• The test audience follows the intended path once with visible ownership.
• A manager can pause the workflow and find the evidence needed to recover it.

Common failures

• The audience or condition is broader than the approved test scope.
• A provider failure leaves work without a human owner.
• The team retries a run without checking whether a side effect completed.

Recovery

• Pause the affected path and inspect the execution evidence before retrying.
• Retry only safe work with the same logical identifier and route failed work to a human owner.

Rollback

• Pause new enrollment and restore the last approved version or fallback workflow.
• Preserve run evidence and reconcile any provider side effects before resuming.

Next task

Read the monitoring and recovery path before expanding the audience.

Open next guide

Automation is useful when the next step is predictable and the team can still understand the state of every run. Design the journey around explicit ownership, timing, and failure handling.

Build from the outcome backward

1. Name the customer or operational outcome first.
2. Define the entry condition and the exact audience.
3. Add one action at a time, with a clear owner or system boundary.
4. Define what happens when an action is delayed, rejected, or unavailable.
5. Run a bounded test before launching to a larger audience.

Campaigns are an operating decision

A campaign is not just a message list. It combines audience, content, channel, timing, and launch controls. Review each of those pieces before starting or pausing a campaign.

Read the run, not only the summary

• A completed step should have evidence behind it.
• A failed step should remain visible and actionable.
• A paused journey should not silently continue through a different path.
• Retries must preserve the original customer and workflow identity.