Here's the short version, because it's the question every founder actually asks us: migrating from Zapier or Make to self-hosted n8n cuts automation costs 70–90% and removes task quotas entirely — but no workflow transfers automatically, so the migration itself is where the real work (and the real savings) live.
We've rebuilt more than 200 workflows for clients moving off Zapier and Make, and the pattern almost never changes. The bill that was fine at 5,000 tasks a month becomes a problem at 80,000. Someone runs the numbers, sees a self-hosted n8n instance costs a flat $20–80 a month regardless of volume, and asks us to move everything. Then they find out there's no import button. This is the guide we wish those clients had read first.
Why Teams Leave Zapier and Make in 2026
The trigger is almost always the bill, and the bill is almost always about how these tools count usage.
Zapier charges per task. Make charges per module step — so a scenario with ten modules running a thousand times burns ten thousand operations, not one thousand. n8n charges per workflow execution, and on the self-hosted plan it doesn't meter executions at all. That single difference in the pricing model is why complex, high-volume automations get 10–20x cheaper on n8n.
The concrete version: a 50,000-task-a-month workload that costs around $1,200/month on Zapier runs comfortably on a $79/month self-hosted n8n VPS. That's north of $13,000 a year, and it stops scaling with your growth. The cost curve flattens.
Money isn't the only reason, though. Teams also move for:
- Self-hosting and data control — everything runs on your own infrastructure, which matters for compliance and for clients who won't allow their data on a third-party SaaS.
- Real code when you need it — n8n's Code node gives you full JavaScript and JSON access, so you stop fighting the no-code ceiling.
- Native AI agents — n8n ships LangChain-based nodes for AI agents and RAG, which is exactly why it's become the backbone for the AI agent systems we build for agencies.
The Hard Truth: Nothing Migrates Automatically
Let's kill the hope right now. There is no reliable one-click converter that turns a Zap or a Make scenario into a working n8n workflow. Every automation gets rebuilt by hand.
I know that sounds like a dealbreaker. It isn't, and here's why: Zapier and Make force you into linear, one-step-after-another chains. n8n doesn't. A node canvas lets a single workflow branch, loop, merge, and call sub-workflows. So the rebuild almost always collapses work. We regularly take three or four separate linear Zaps — the classic "when this, then also do this other thing" sprawl — and consolidate them into one clean n8n workflow that's easier to read and cheaper to run.
The rebuild isn't a tax. It's a forced cleanup of automation debt you've been carrying for years.
The Migration Process We Actually Run
Five phases. It's deliberately unglamorous, because every failed migration I've seen skipped the boring parts.
1. Audit before you touch anything
Open a spreadsheet. List every active Zap or scenario with four columns: trigger, actions, monthly task volume, and business criticality. This one document does most of the strategic work — it surfaces the handful of expensive, high-volume workflows that are worth migrating first and the long tail of tiny automations you can leave alone or delete entirely. Half the "workflows" in a mature Zapier account turn out to be abandoned.
2. Stand up infrastructure properly
Before rebuilding a single workflow, deploy n8n on a managed VPS with a real database, automated backups, an error-trigger workflow that catches failures, and basic monitoring. This is the step hobbyists skip and then blame n8n when a workflow silently dies at 2 a.m. It's not n8n that's unreliable — it's an unmanaged box with no backups. Treat it like production, because it is.
3. Rebuild highest-value workflows first
Start with the automations from your audit that carry the most volume and the most business risk. Recreate them natively — don't try to photocopy the Zapier structure node-for-node. Ask "what should this do?" not "how did Zapier do this?" That's where the consolidation wins come from.
4. Run both platforms in parallel
This is the step that separates a clean migration from an incident. For each rebuilt workflow, keep the original running and run the n8n version beside it for three to seven days, feeding both the same inputs and comparing outputs. Silent failures — a field that maps slightly differently, a date format that shifts — show up here instead of in a client's inbox.
5. Cut over, then decommission
Once a workflow has run clean in parallel and matched outputs, switch the live trigger to n8n. After everything has been stable on n8n for at least two weeks, downgrade or cancel the old plan. If you want a safety net, Zapier's free tier keeps 5 Zaps alive at no cost.
Zapier vs Make vs n8n: The Quick Comparison
| Factor | Zapier | Make | Self-hosted n8n | |---|---|---|---| | Billing model | Per task | Per module step | Per execution (unmetered self-hosted) | | Typical monthly cost at scale | $$$ | $$ | $ (flat VPS cost) | | Ease of setup | Highest | High | Requires infrastructure | | Custom code / JSON | Limited | Partial | Full (Code node) | | Self-hosting / data control | No | No | Yes | | Native AI agents / RAG | Add-ons | Add-ons | Built in (LangChain) | | Best fit | Simple, low-volume | Mid-complexity, visual | Complex, high-volume, AI-heavy |
The rule of thumb: if your automations are simple and low-frequency, Zapier's convenience can be worth the premium. The moment you're running complex, high-volume, or AI-driven workflows — especially across multiple clients — n8n wins on both cost and capability.
The Traps That Break Migrations
A few things go wrong often enough that they're worth calling out:
- Migrating everything at once. Move the expensive, high-value workflows first. Leave the low-risk long tail until the core is stable. A big-bang cutover is how you turn a cost-saving project into an outage.
- Skipping the parallel-run. If you flip the switch without running both versions side by side, you will miss a mapping difference, and you'll find it when a client does.
- No error handling. Zapier surfaces failures in its UI. On self-hosted n8n, if you don't build an error-trigger workflow and alerting, a broken automation just goes quiet. Build the safety net before you need it.
- Under-provisioned infrastructure. A $5 box with no backups is not where your revenue-critical automations should live. The hosting is the cheapest part of the whole thing — don't cheap out on it.
We wrote more about the workflows worth building first in the n8n workflows every agency needs, and if attribution is part of your stack, our n8n Hyros nodes plug straight into a migrated setup.
Should You Migrate Yourself or Hire It Out?
If you're technical, have fewer than a dozen workflows, and enjoy this stuff, a DIY migration over a couple of weekends is completely reasonable — the official n8n docs are genuinely good. Where teams call us is when the account is large, the workflows are business-critical, or nobody internally wants to own self-hosted infrastructure and its backups. In those cases the migration pays for itself inside the first quarter, and you get a cleaner, cheaper, more capable automation layer on the other side — plus room to bolt AI agents on top of it.
Get a Free Automation Audit. We'll map your current Zapier or Make workflows, estimate exactly what self-hosted n8n would cost you, and show you the consolidation opportunities before you commit to anything. Start your free audit — it's the same audit spreadsheet we described in step one, run on your actual stack.