Zapier to n8n Migration Guide

One automation agency I worked with was paying $847/month to Zapier for workflows that cost them $31/month to run on their own n8n instance. That's a 96% cost reduction for the same automations, plus capabilities Zapier literally cannot offer.

Quick on who's talking:

• AI ops operator for service firms , 40+ systems deployed for service businesses
• Built n8n + Claude workflows that have automated 6,000+ hours of repeat work
• Work spans 8+ industries (construction, manufacturing, staffing, property management, agencies)
• Systems run on self-hosted n8n + Supabase, not Zapier glue

Why This Migration Matters Right Now

Zapier is excellent at the first 20 automations of a business. It's terrible at the next 200. Once your operation scales past a handful of zaps, three problems compound: per-task pricing that punishes growth, flexibility ceilings that force ugly workarounds, and a complete inability to run real AI agent workflows without bolting on six other tools.

The numbers I see in client audits are consistent. Service businesses between 11 and 100 employees typically run 40-120 active zaps and pay between $300 and $1,200 per month. That same workload on a self-hosted n8n instance costs $20-60/month in infrastructure. The migration takes 2-6 weeks depending on complexity. The payback period is almost always under 90 days.

If your Zapier bill has crossed $300/month, you are no longer "saving time with automation." You are renting workflows you could own outright.

1. Audit Your Current Zapier Setup

Before you migrate a single zap, you need a complete inventory. Most teams skip this step and pay for it later when a "small" zap turns out to be the linchpin holding invoicing together.

Step 1: Export your zap inventory

Open your Zapier dashboard and document every active zap. For each one, capture trigger, actions, monthly task volume, and business criticality. A spreadsheet is fine , what matters is that nothing gets missed.

Zap Name	Trigger	Actions	Tasks/Month	Criticality
New Lead to Slack	HubSpot	Slack, Sheets	~200	High
Invoice Created	Xero	Gmail, Notion	~50	High
Form to CRM	Typeform	HubSpot	~300	Medium
Stripe Receipt Archive	Stripe	Drive, Airtable	~120	Low

Step 2: Identify your core integrations

Look for the apps that show up across 60%+ of your zaps. Those are your core integrations and they need to work flawlessly in n8n before you migrate anything else.

• CRM: HubSpot, Salesforce, Pipedrive , all have first-class n8n nodes
• Accounting: Xero, QuickBooks , supported, sometimes via HTTP node for newer endpoints
• Communication: Slack, Gmail, Microsoft Teams , native nodes, full feature parity
• Project management: ClickUp, Notion, Asana , solid nodes, occasional gaps on niche endpoints
• Forms: Typeform, Tally, Google Forms , webhook-based, more reliable than Zapier's polling

Step 3: Flag the complex ones

Some zaps migrate in 15 minutes. Others take a full day. Flag any zap that uses Zapier Paths, Formatter, or Filter heavily, has more than 5 steps, connects to apps with no obvious n8n equivalent, or sits in the middle of a revenue-critical process like invoicing or onboarding.

2. Score and Prioritize What to Migrate First

Not every zap deserves migration. Some should be deleted. Some should be rebuilt from scratch because they were bad to begin with. A simple scoring rubric saves weeks.

Criterion	1 point	3 points	5 points
Monthly task volume	Under 50	50-500	500+
Business criticality	Nice to have	Important	Revenue-critical
Zapier cost driver	Trivial	Moderate	Top 5 in account
Complexity to rebuild	Hours	Half a day	1-3 days

Migrate high-volume, high-criticality, high-cost zaps first. Defer or delete the long tail. In a typical 80-zap account, roughly 20 zaps drive 80% of the cost , migrate those and you've already justified the entire project.

The migration is not "move every zap." The migration is "move the 20% of zaps generating 80% of your bill, then decide what's worth keeping at all."

3. Stand Up Your n8n Instance Properly

You have two real options: n8n Cloud or self-hosted. The right choice depends on your team, your data, and your appetite for owning infrastructure.

n8n Cloud vs self-hosted

Factor	n8n Cloud	Self-hosted
Monthly cost (mid-volume)	$50-200	$15-40 (VPS)
Setup time	10 minutes	2-4 hours
Data residency control	Limited	Full
Execution limits	Per-plan caps	Hardware-bound
Maintenance burden	None	Updates, backups, monitoring
Best for	Teams without DevOps	Teams with any technical capacity

For most service businesses in the 11-100 employee range, I recommend self-hosted on a $20-40/month VPS (Hetzner, DigitalOcean, Railway). The cost difference compounds, and self-hosted unlocks the ability to run sensitive data workflows without sending it through anyone else's infrastructure.

The setup checklist

1. Provision a VPS with at least 4GB RAM and 2 vCPUs (8GB if you'll run AI workflows)
2. Install n8n via Docker Compose with a persistent Postgres database, not SQLite
3. Set up a reverse proxy (Caddy or Nginx) with automatic HTTPS
4. Configure automated backups of the Postgres database to an off-server location daily
5. Enable basic auth or SSO before you put a single credential in
6. Set up an error workflow , every workflow you build will reference this one for failure alerts

4. Map Zapier Concepts to n8n Equivalents

Most of the friction in migration comes from translating Zapier's mental model to n8n's. They look similar but think differently.

Zapier concept	n8n equivalent	Practical difference
Zap	Workflow	n8n workflows can have multiple triggers and branches
Trigger	Trigger node	Webhook triggers are first-class and instant
Action	Regular node	Each node returns structured data you can reuse anywhere downstream
Paths	Switch / IF node	Cleaner branching, no limit on path count
Formatter	Set / Code / Edit Fields	Full JavaScript available when you need it
Filter	IF node or Filter node	Same idea, more readable conditions
Storage by Zapier	Postgres / Redis / Supabase	Real database, real queries, no row caps
Webhooks by Zapier	Webhook trigger	Built in, no extra cost, no task consumption
Sub-zap	Execute Workflow node	Sub-workflows are reusable and testable independently

The biggest shift: in Zapier, data flows in a straight line. In n8n, data flows as items through nodes, and any downstream node can reach back to any upstream node's output. This is why complex workflows that took 12 steps in Zapier often collapse to 5 nodes in n8n.

5. Migrate Your First Five Workflows the Right Way

Don't start with the hardest workflow. Start with one that's medium complexity, medium criticality, and well-documented. You're learning the platform, not proving anything.

The migration loop

1. Document the zap end-to-end , trigger details, every field mapping, every filter condition, every formatter operation
2. Build the n8n workflow in parallel , do not turn off the zap yet
3. Use test data in n8n until outputs match the Zapier outputs node-for-node
4. Run both in parallel for 48-72 hours, comparing results daily
5. Cut over by disabling the Zapier zap and routing traffic to the n8n workflow
6. Leave the Zapier zap paused (not deleted) for 2 weeks in case you need to revert

Common migration gotchas

• Date formats: Zapier silently coerces dates. n8n is stricter , you'll need explicit ISO 8601 formatting in many places.
• Empty fields: Zapier skips empty fields by default. n8n sends them as null or empty strings unless you handle it.
• Rate limits: Zapier throttles you invisibly. n8n will hammer an API at whatever speed it can , use the "Limit" or "Wait" nodes for rate-sensitive integrations.
• Error handling: Zapier emails you when a zap fails. In n8n, you must explicitly wire up an error workflow or you'll never know things broke.
• Polling vs webhooks: Many Zapier triggers poll every 1-15 minutes. n8n webhook triggers fire instantly , your workflows often feel faster overnight.

Run old and new in parallel. Always. Every team I've watched skip the parallel-run phase has eaten a production incident inside the first 30 days.

6. Handle the Hard Cases

About 15-20% of zaps in a mature account will not have a clean 1:1 migration. Here's how to handle the patterns that come up most.

Zaps using Formatter heavily

Zapier Formatter is a black box. In n8n, you'll rewrite these using the Set node for simple transforms, the Code node for anything involving loops, regex, or conditional formatting. The Code node accepts plain JavaScript , most Formatter operations become 3-10 lines of code that you can actually read and debug.

Multi-step Paths with deep branching

Zapier Paths max out at 3 branches per level and get unwieldy past that. In n8n, the Switch node handles any number of branches cleanly, and you can nest IF nodes for compound conditions. A 14-path zap I migrated last year became a single Switch node with 14 outputs and a fraction of the maintenance burden.

Apps with no native n8n node

Use the HTTP Request node. Any app with a REST API works , and that's almost all of them. The HTTP node supports OAuth2, API keys, headers, query params, and JSON bodies. For apps you'll call repeatedly, save the credential once and reuse it across workflows.

Workflows that touch sensitive data

This is where self-hosted n8n earns its keep. Customer PII, financial data, health information , none of it has to leave your infrastructure. For one staffing client, moving candidate data workflows from Zapier to self-hosted n8n removed three vendors from their data processing inventory and simplified their SOC 2 review.

Workflows that need AI agents

Zapier's AI features are limited and expensive per task. n8n has a full AI Agent node that integrates with Claude, OpenAI, and any other LLM provider, plus vector stores like Supabase and Pinecone. If you're planning any AI workflow more sophisticated than "summarize this email," n8n is genuinely the only practical option in this price tier.

7. Decommission Zapier Without Breaking Anything

The final phase is the one most teams botch. After 4-6 weeks of migration, you'll have 80% of traffic running on n8n and a Zapier account still costing real money. Closing it out cleanly matters.

The decommission sequence

1. Audit what's still active in Zapier , anything still running after week 6 is either critical-and-stuck or forgotten
2. Make a kill/keep/migrate decision for each remaining zap with the actual workflow owner
3. Migrate the keepers as a final batch
4. Pause (don't delete) the kill list for 30 days to confirm nothing breaks
5. Downgrade your Zapier plan to free or starter the moment you're under the threshold
6. Cancel after 60 days of full n8n operation with no incidents

What to expect from the numbers

Metric	Before (Zapier)	After (n8n)
Monthly platform cost	$300-1,200	$20-60
Average workflow execution time	2-15 min (polling)	Under 5 sec (webhooks)
Custom logic capability	Limited	Unlimited (full JS)
AI workflow support	Minimal, costly	Native, cheap
Vendor data exposure	Every workflow	Zero (self-hosted)
Time to build a new workflow	30-90 min	20-60 min after learning curve

The cost savings are the easy part to quantify. The harder-to-measure wins are bigger: workflows that used to be impossible become trivial, sensitive data stays in-house, and your team stops architecting around platform limits.

What to Do This Week

Pick one afternoon. Open your Zapier dashboard. Export the zap inventory using the template in Section 1. Score them with the rubric in Section 2. By the end of that session, you'll know exactly what your migration looks like, how long it will take, and what your new monthly bill will be.

If you'd rather not run the migration yourself, this is exactly the kind of project I run for service businesses between 11 and 100 employees. A typical migration takes 3-5 weeks, costs less than three months of the Zapier bill it replaces, and ships with documentation your team can actually maintain. If that's the path you'd rather take, the link below opens a strategy call where we'll map your current setup and price the move.