Migration · 5 Nov 2026 · 8 min read
How to Migrate from Zapier or Make to n8n (Step by Step, With Pitfalls)
A practical guide to migrating from Zapier or Make to n8n: audit, concept mapping, rebuild, parallel run, cutover, and the pitfalls that cause silent failures.

Short answer: there is no one click export from Zapier or Make into n8n. A safe migration is a rebuild in five phases: audit, map, rebuild, run in parallel, cut over. Expect to recreate logic and reconnect credentials, and treat the parallel run as non negotiable.
Teams usually switch for one of three reasons. The bill grows faster than the value, the logic outgrows the tool, or someone wants control over where data lives. If you haven’t checked whether the numbers justify a move, start with n8n vs Zapier: cost per 1,000 finished jobs.
Disclosure: AnnamNode runs n8n migrations as a service. This guide is written so you can do the work yourself if you prefer.
Should you migrate at all?
Move when at least one of these is true. Your task or operations bill climbs faster than the value of the automations. Workflows need loops, branching, custom code or AI steps that feel strained. You need self hosting, data residency or tighter control of credentials. Or you want version control, proper environments and real error handling.
Stay, or migrate selectively, when your workflows are short, stable and low volume, when you rely on niche apps that only have a Zapier or Make connector, or when nobody on your side will own an n8n instance. Plenty of teams end up with a hybrid: the simple Zaps stay where they are and the heavy workflows move.
What does not transfer
The platforms are built differently, so you can’t export a Zap and import it. Zapier lets you share Zaps as templates, not as portable workflow files. Third party converters exist (for example migromat converts Zapier and Make exports toward n8n) and they can speed up a first draft, but converted output still needs review and testing. Treat a converter as a head start. It is not a migration.
Phase 1: audit what you have
Make an inventory of every Zap or scenario. For each, write down its name and owner, the trigger type, the apps it touches, steps and branches, monthly runs or tasks, the business impact if it fails, and when it was last edited. Then sort the list into three piles: retire what nobody uses, migrate now what is expensive or complex, and leave what is simple and cheap.
Image: Pixabay (free to use)
Phase 2: map concepts and check coverage
Most ideas have a direct cousin in n8n. A Zap or scenario becomes a workflow, a module becomes a node, and Paths or a Router become IF and Switch nodes with Merge to rejoin them. Filters become an IF node. Formatter and text tools become the Set node, expressions or the Code node. Storage becomes data tables, a database or Google Sheets, and replay or error handling becomes an error workflow with the Stop and Error node and per node retry. Connections are called credentials.
Two differences catch people out. First, n8n pushes lists of items through each node, while Zapier typically runs an action once per record and bills each time, so a loop over ten rows is a single execution in n8n. Second, data mapping is more explicit: you reference earlier nodes by expression instead of picking from a dropdown, and a little JSON literacy goes a long way.
Check coverage honestly. Zapier lists 8,000+ integrations and n8n lists 1,000+, plus the HTTP Request node for any REST API, GraphQL and webhooks. If an app has no node, see whether it has an API, and if it doesn’t, mark that workflow “leave in place”.
Phase 3: rebuild in priority order
Pick one pilot, a medium complexity workflow with a measurable outcome. Set up the environment, either n8n Cloud or self hosted (how to decide), and if you self host, get backups and monitoring in place first. Then reconnect credentials inside n8n with least privilege access. Don’t paste secrets into notes or chat, because credentials must be recreated and they do not migrate.
Rebuild the logic from the trigger outward, using the HTTP Request node where no native node fits. Add what the old tool never gave you: an error workflow, alerts to Slack, email or Telegram, naming conventions and notes on each node. Finish with a paragraph of documentation covering what it does, its trigger, its owner and how to roll back. Build the error handling during the rebuild and not afterwards; n8n error handling and the n8n production checklist list what to wire up.
Image: Pixabay (free to use)
Phase 4: run in parallel
Run old and new side by side before you switch. Feed them the same inputs for a defined window, say one to two weeks of real traffic, and compare outputs. The big danger is double action: if both workflows send emails or create CRM records, point the new one at a test destination or gate its output so a customer never gets two messages. Log every difference, whether a missing field, timing or formatting, fix it, and repeat until each difference is explained.
Phase 5: cut over and retire
Choose a quiet window. Switch the trigger source, for example by changing the webhook URL in the source app. Disable the old Zap or scenario but don’t delete it yet. Watch the first runs closely with alerts on. After a stable period, at least a few weeks for anything business critical, delete the old automation and cancel the plan tier you no longer need. Keep a rollback note that says how to switch the old workflow back on within minutes.
Pitfalls we keep seeing
Lift and shift thinking is the classic one: rebuilding a Zap node for node and missing simpler n8n patterns, so redesign around items and branching. Polling versus webhooks is next. If the old workflow used a webhook and the new one polls, execution counts balloon, so use webhook triggers where possible. Execution time limits bite on Cloud (5 minutes on Starter and 40 on Pro per n8n’s pricing), so split long jobs. Duplicate actions during the parallel run, missing error paths, rate limits (n8n loops faster than the old tool, so add waits and batch sizes), over broad API keys copied across, and the quiet killer of all migrations: no owner. Name one, write it down, schedule reviews.
How long does it take?
It depends on how many workflows you have and how tangled they are, which is why the audit comes first. One straightforward workflow can be rebuilt quickly. A portfolio of dozens that share logic is a project with a plan. In our discovery sprint we map your automations, rank them by return and give a fixed quote for the build. If you are budgeting, see what an n8n agency costs.
A migration checklist
- Inventory of every Zap or scenario with run counts
- Retire, migrate or leave decided for each
- Connector coverage checked, HTTP fallbacks identified
- n8n environment, backups and monitoring ready
- Credentials recreated with least privilege
- Error workflow and alerts attached to each migrated workflow
- Parallel run finished with every difference explained
- Cutover plan and rollback note written
- Old automations disabled, then deleted after a stable period
- ROI measured against the baseline (see automation ROI)
FAQ
Can I export Zapier workflows to n8n?
Not directly. The architectures differ, so you rebuild. Converters can produce a first draft, but review and test everything.
Can I migrate from Make to n8n?
Yes, with the same five phases. Make scenarios map well because both tools use a visual canvas with routers and branches, though module configuration and data mapping differ.
How do I avoid downtime?
Run old and new in parallel, gate outputs, cut over in a quiet window and keep the old workflow disabled but available for rollback.
Will my costs go down?
For multi step or high volume workflows, usually, once you include build effort and hosting. Compare using cost per 1,000 finished jobs.
Do I need to self host?
No. n8n Cloud avoids infrastructure work. Self hosting is a choice for control and data residency.
Related reading
- n8n vs Zapier: cost per 1,000 finished jobs
- n8n pricing 2026: cloud vs self hosting
- n8n error handling: why workflows fail silently
- How much does an n8n automation agency cost?
Sources
- n8n vs Zapier comparison (vendor)
- n8n integrations
- n8n HTTP Request node
- n8n plans and pricing
- n8n community: Zapier to n8n discussions
- migromat (third party converter, listed as an example, not an endorsement)