How to run old and new automation side by side before you switch

A parallel run means both copies see the same trigger window. Only one job. Not every automation you have. Keep the old job on. Rebuild that job on the new tool. Compare the writes. Then turn the old job off.

Switching every job on the same day hides the first miss. The new tool may skip a row, send a duplicate Slack, or fail a connection you only notice after the old job is off. Catch that miss while the old job can still write.

You cannot paste one tool’s export into another and expect a working job. Auth, history, and steps that only exist in one tool stay on the old tool — see what does not migrate — this page does not reprint that list. A green history icon is not the write. Catch that on when automations look fine and name the class on the failure taxonomy before you treat the new job as done.

Pick one job you can compare. Leave the rest until this one matches.

• The job whose miss is noticed first. A lead write, an invoice ping, a support alert. If nobody would notice a skip, leave it manual — see what’s worth automating.
• A write you can open and match. Sheet row, CRM record, Slack message. You need a destination you can check field by field, not only a vendor checkmark.
• The failure path, not only the happy path. Force a missing required field and an expired connection once. Name the class on the failure taxonomy — auth, bad data, and expected non-match are different jobs.
• Not a live charge, invoice book, or one-shot customer email. Rebuild those with the new write off, or write to a sandbox, until outcomes match. Do not invent a monthly dollar from the side-by-side window.

Name who compares the two copies and who can turn the old job back on — who should own your automations. Recreate connections on the new tool. They do not travel.

A test click is not success. Compare a real trigger window. All of these must hold:

• Same write. Same row, record, or message fields — including empty fields you meant to leave empty.
• Same skip. A non-match that the old job leaves alone, the new job also leaves alone.
• Same failure class. When you expire a connection or omit a required field, both copies fail the same way. Name the class first: when automations fail. Do not trust a green history icon — when automations look fine.
• Units compared separately. Do not treat one task as one credit as one execution. No converted monthly dollar.
• Owner confirmed the new write. Someone named opened the sheet, CRM, or inbox — not only the run list.

Do not switch on the same day as the first rebuild. Pass every item before you turn the old job off:

• A real run matched the old job, not only a manual test.
• New-tool connections were recreated and tested. They do not migrate — what does not migrate.
• Someone is named for reconnects and 2am — who should own your automations.
• The way back is written: the old job still exists and can be turned back on.
• Steps that only exist in one tool, and org permissions, were rebuilt — not assumed from an export.

Still choosing the new tool? Four questions live on which automation tool for small teams — this page does not reprint that flowchart.

Going back is turning the old job back on. It is not a vendor “undo migrate” button. You cannot reverse a cross-tool move with one click.

• Keep the old job intact until the new job has handled a real run.
• If the new job misses or duplicates, turn the new job off. Leave the old job on.
• If you already turned the old job off, turn the old job back on before you debug the new one.
• Do not delete the old export, connections, or run history on switch day. History does not travel.
• Name who can flip the switch before go-live. An orphaned job has no way back — who should own your automations.

While both copies run, each vendor counts a different thing. A Zapier task is not a Make credit and is not an n8n execution. Do not treat one task as one credit as one execution. Full unit math lives on How to meter automation and Same job, three meters — this page does not reprint those tables. This page does not invent an overage dollar or a monthly dollar.

Both copies matched, or you skipped the side-by-side. Use the sibling that owns the next job — without treating this page as a ranking.

• Still deciding if the job is ready: What’s worth automating — and what to leave manual.
• What stays behind in an export: What does not migrate between Zapier, Make, and n8n — this page does not reprint that list.
• Units before a plan: How to meter automation · Same job, three meters — tasks vs credits vs executions.
• When a run fails or still looks on: When automations fail — name the class first · When automations look fine — catch silent breakage · Error handling across Make, n8n, and Zapier.
• Nobody is named for reconnects or 2am: Who should own your automations.
• Four questions, then a pick: Which automation tool for small teams.
• When a named product is the job: When Make is the right pick (and when it is not) · When Zapier is still the right pick (and when it is not) · When n8n self-host is worth it (and when it is not).
• When to pick each: Make vs n8n vs Zapier — when to pick each · live scorecard (totals are not published yet; self-host is a separate lane).

Then pick a tool

Primary CTA stays the live scorecard. This page does not reprint the picker flowchart.

Primary vendor export and import pages. Access date for every export note on this page: 2026-09-10. This page does not publish list prices. We do not invent a cross-vendor import path, a converted list, or an overage dollar.

• Make — blueprints · accessed 2026-09-10
• n8n — export and import · accessed 2026-09-10
• Zapier — import and export Zap workflows · accessed 2026-09-10