How do I test an automation before publishing?

Last updated: August 26, 2026

Save your changes as a draft, run the automation against a real record to confirm it behaves the way you expect, then publish — and roll back if it doesn't.

Editing a live automation is risky when you can't see what it'll do until it fires on its own. Onboarded gives you a way to make changes, test them against a real record, and only then make them live. This guide walks through that loop.

Before you start

  • An automation you've already built. If you're starting from scratch, build it first.

  • The ID of a real record to test against — a placement or task that already exists in your account.

How versions work

Every automation keeps a version history. When you open a published automation and start editing, your changes go into a new draft version. The published version keeps running untouched until you publish the draft — so editing never disrupts what's live. The version selector at the top of the editor shows each version by number (v1, v2, and so on), and the status badge tells you whether you're looking at the Published version or a Draft.

Save your changes as a draft

Make your edits in the editor. Field changes auto-save as you go, but click Save draft to lock in the current state before you test. This keeps your draft separate from the published version and gives the test run a stable version to execute.

Run the automation against an existing record

A test run targets a placement or task that's already in your account. Onboarded doesn't create a new placement or task for the run, so you don't have to set up fresh test data every time — point the run at a record you already have and reuse it as often as you need.

  1. Click Run in the editor's top navigation. The Run Automation panel opens.

  2. Enter the ID of the record you want to trigger the automation with. The panel takes a placement ID (prefixed pla_) or a task ID (prefixed task_), and the prefix has to match the automation's trigger entity — the field is labelled Placement ID or Task ID depending on which entity the trigger uses, and a task_ ID won't work on a placement-triggered automation. The panel loads a preview of the trigger event for that record.
    You should see the preview populate with the record's data.

  3. Check the preview for any missing paths. It lists every field the automation references alongside the value it resolved to for this record, and flags anything that doesn't resolve with an amber warning icon and the words Not set. The footer tallies how many fields are missing and warns that the automation may not behave as expected — those are the fields an action would send empty or fail on.

  4. Click Run to execute the automation against that record.

If you have unsaved changes, you'll be prompted with Save draft before running? — save first so the run executes your latest edits rather than stale data.

A manual run performs the automation's real actions. If the automation sends an email, running it sends a real email to the record's employee. Test with a record you control — or in a UAT account — so a live employee doesn't receive a test message. If you enter an ID that doesn't exist, the preview returns Trigger object not found; double-check the ID and the trigger entity match.

Publish when it's ready

Once the test run does what you expect, click Publish. Your draft becomes the live version, the status badge changes to Published, and the automation now fires on real events.

Roll back if something's off

If a draft isn't working out before you've published it, click Delete draft to revert to the previous version. Because every version is kept in the version selector, you can also open an earlier version to compare against your current one.

Deleting a draft can't be undone. It discards your in-progress changes and restores the previous version. Confirm you don't need the draft's edits before deleting it.

Verify

After a test run, open the automation and check the Events column for a successful run count. For a block-by-block view of the run — the logs, the data each block evaluated, and a Rerun button — see 📄 How do I check an automation's run history with the Events panel? If the run errored, the most common causes are a missing path flagged in the preview, a trigger entity that doesn't match the record type, or an action pointed at an integration that isn't connected — such as an older email action still set to a SendGrid account.

Related