'Zapier or Make Is Getting Expensive: A Migration Guide to n8n'

Your Zapier invoice grows every month even though you built nothing new. That is no accident. That is the pricing model.

I did this move for our own brands and then for several Flowhouse clients. It is very doable. But it goes wrong when you treat it as a weekend project and switch everything at once.

Why the invoice grows

Zapier and Make bill per task or per operation. Every single step in your process counts. A workflow with eight steps therefore burns 8,000 units over 1,000 runs, not 1,000.

That is the point where many people get surprised. Your orders double, your automation bill quadruples, because in the meantime you also added two steps per workflow.

On top of that come soft limits that only hurt in daily operation:

• Runtime per step. Slow API responses get cut off.
• Data volume. Large lists have to be split artificially.
• Logic. Branches and loops are possible, and they cost extra tasks every time.
• Export. Your process logic stays with the vendor. You cannot take it with you.

With n8n you either pay for the server or for a cloud plan billed by executions. A workflow with eight steps costs the same there as one with two. The full comparison of the three tools is in n8n vs Make vs Zapier.

When the move actually pays off

Not always. My threshold looks like this.

Moving pays off when at least two of these apply:

• You pay more than 150 euros a month for automation.
• Your workflows have more than five steps.
• You want to add your own code, for a calculation or your own data source.
• You have requirements about server location or data sovereignty.
• A process has become business critical, order processing for instance.

Stay where you are when:

• You pay under 50 euros a month. The move costs you more time than it saves.
• Nobody on the team wants to look after a server and n8n Cloud is too expensive for you.
• Your workflows will look different in three months anyway.

Do the maths once. Moving ten workflows realistically costs 5 to 10 working days. If that saves you 200 euros a month, you have it back within half a year. At 40 euros of savings, never.

What else automation costs beyond the tool invoice is in what automation really costs.

Inventory first, then the move

Before you even install n8n, you need a list. For every existing workflow:

• How often does it run per month?
• How many steps does it have?
• Which systems does it touch?
• What happens if it goes down for a day?
• Is it even still running?

The last point is the most rewarding. In our own inventory, seven out of 23 active Zaps were dead: built for processes that no longer existed. We deleted those instead of migrating them. That is the cheapest part of any migration.

The order

Never migrate by importance. Migrate by risk, from the bottom up.

Wave 1: the harmless ones. Notifications, internal reports, anything without customer contact. If something goes wrong here, only your team notices. And you learn how n8n behaves.

Wave 2: the internal data flows. Pushing data between your own systems, updating tables, reconciling stock. Errors are visible, and they are fixable.

Wave 3: everything with customer contact. Order processing, shipping emails, support replies. Only now, and only with parallel running.

Wave 4: the money. Payments, refunds, bookkeeping. Last, with double checking.

For us wave 1 took two days, wave 3 alone almost a week. The order is what keeps the move calm.

What won't transfer one to one

There is no button that turns a Zap into an n8n workflow. Five things you really have to rebuild.

The built-in conveniences. Zapier's formatting steps, delays and filters are their own blocks. In n8n you build that with a function or an If node. Usually shorter, and different.

Triggers that don't exist. Not every app has a ready-made trigger in n8n. Often you need a webhook instead, an address that the other app calls when an event happens. If that is new to you: webhooks explained.

Credentials. Every connection has to be set up again. Budget half a day just for hunting down API keys and clicking through authorisations.

Error handling. Zapier retries a lot automatically and sends you an email. In n8n you define yourself what happens on an error. That is more work and better in the end, because you decide what even counts as an error.

Data formats. Zapier often hands you fields flat, n8n works with nested objects. Expect to touch every field mapping once.

If you have never used n8n, build one throwaway workflow first. Your first n8n workflow is entirely enough for that.

Parallel running: the only safe test phase

This is the part almost everyone skips, and the reason migrations hurt.

Here is how we do it: the new n8n workflow runs alongside the old one, but idling. It processes the same data and writes its result to a log instead of sending it. No email goes out, no order gets created.

After two weeks you compare. In our order processing this surfaced three discrepancies that would never have appeared in an isolated test: an address with an umlaut, an order with a discount code and a case with two shipping addresses.

Then you switch over, in stages:

1. New workflow live, old workflow off, and not deleted.

2. Ten minutes of checking every day for a week.

3. Only after that, delete the old workflow and downgrade the plan.

Point three gets forgotten a lot. If you don't cancel the old plan, you keep paying, and the whole move was for nothing.

Data protection and GDPR

Switching to n8n is also a data protection decision, and usually a good one.

Self-hosted means your data stays with you. Customer names, addresses and email content no longer run through a vendor's servers in the US. Our n8n runs on a server in Germany. Whether that makes sense for you is covered in n8n self-hosting vs cloud.

You still need contracts. For n8n Cloud you need a data processing agreement, exactly as with Zapier. When self-hosting you need one with your hoster.

Update your record of processing activities. If Zapier is listed there and Zapier is gone, the line needs changing. It takes ten minutes and it is the first thing anyone looks at in an audit.

Clean up the logs. By default n8n stores every run including its content. With support emails that is real customer data. Set the retention deliberately; ours is 30 days.

Revoke access after the move. The old vendor still holds authorisations for your Shopify, your inbox, your warehouse. Revoke them actively, it doesn't happen on its own.

The limits

What the move does not solve:

• Bad workflows stay bad. If your logic is unclear, in n8n it will just be unclear more cheaply.
• You take on operations. Self-hosted means updates, backups, uptime. That is a few hours per quarter, and it is more than zero.
• Fewer ready-made blocks. Not every niche app has a node. A direct API call usually solves it, and you have to read the docs for that.
• The move costs real days. If you don't have 5 to 10 working days, better to postpone it than to do it halfway.

Conclusion

The move to n8n pays off when your invoice has three digits and your workflows have become serious. It does not pay off as a matter of principle.

The three things that make the difference: a thorough inventory beforehand, the order from harmless to critical, and two weeks of parallel running before you switch.

If you have your Zapier invoice in front of you and want to know whether the effort adds up: write to us via Flowhouse. We will also tell you when it doesn't.