How to Migrate from Zapier to n8n in 2026

Zapier is the easiest tool to start with, and the hardest to stay on once costs cross $500/month. The n8n 2.0 release in January 2026 made migration more attractive than ever — native AI agents, 500+ deep integrations, and self-hosting that cuts costs 60–90%. Here’s the exact playbook teams are using in April 2026.

Last verified: April 21, 2026

Why migrate at all?

The three triggers that push teams from Zapier to n8n in 2026:

1. Cost. Zapier’s task-based pricing scales linearly. A 10-step zap running 1,000 times/day = 10,000 tasks/day = $500+/month on the Pro plan alone. Same workflow on self-hosted n8n: ~$20/month total.
2. AI agents. n8n 2.0 ships 70+ AI nodes, LangChain integration, persistent memory, self-hosted LLM support, and an MCP client node (April 2026). Zapier’s AI is shallower.
3. Control and portability. Self-hosted n8n means your automation layer is portable, forkable, and auditable. Zapier is a black box you can’t export.

Before you start: decide cloud vs self-hosted

Option	Best for	Cost
n8n Cloud Starter	0–2K executions/mo	$20/mo
n8n Cloud Pro	2K–10K executions/mo	$50/mo
Self-hosted (Hetzner/DO)	Unlimited, you manage	$5–40/mo hosting
Self-hosted (k8s/enterprise)	Regulated industries	$100+/mo infra

For most teams migrating from Zapier Pro ($49/mo) or Business ($103/mo), self-hosted n8n is the obvious upgrade — the hosting is cheaper than your old Zapier bill, and you get unlimited executions.

Step 1 — Inventory your zaps (1–2 hours)

Export your Zap list. In Zapier:

1. Go to My Zaps → All Zaps
2. Filter by Status: Active (ignore drafts and paused zaps)
3. Export or screenshot the list
4. For each active zap, note:
   • Trigger app and event
   • Number of action steps
   • Apps used in actions
   • Tasks per month (from Zap History)

Output: a spreadsheet with one row per zap, sorted by monthly task volume (highest first).

Step 2 — Apply the 80/20 rule

Typically, 20% of your zaps drive 80% of your task cost. Migrate those first. The long tail can wait — and some of them are probably candidates for deletion.

In practice:

• Top 20% (high-volume): migrate immediately
• Middle 60%: migrate in week 2
• Bottom 20% (low-volume, rarely-triggered): audit — often you can delete them

Step 3 — Set up n8n (30 minutes)

Option A: n8n Cloud (easiest)

1. Sign up at n8n.cloud
2. Pick the Starter plan ($20/mo) or the 14-day Pro trial
3. You’re done. Skip to Step 4.

Option B: Self-hosted on a VPS (best cost)

On a Hetzner CX22 ($4.59/mo) or DigitalOcean $6 droplet:

# Install Docker
curl -fsSL https://get.docker.com | sh

# Run n8n
docker run -d \
  --name n8n \
  -p 5678:5678 \
  -v n8n_data:/home/node/.n8n \
  -e N8N_HOST=your-domain.com \
  -e WEBHOOK_URL=https://your-domain.com/ \
  -e GENERIC_TIMEZONE=Europe/Tallinn \
  --restart always \
  n8nio/n8n

Add Caddy or Nginx for HTTPS with auto-SSL. You’re live in 15 minutes.

Option C: n8n Railway template (middle ground)

# One-click deploy via Railway template
https://railway.app/template/n8n

Railway handles TLS and scaling; costs ~$5–20/month depending on usage.

Step 4 — Rebuild your top zap

Pick your highest-volume zap. Open it in Zapier and in n8n side-by-side.

The 1:1 mapping

Zapier concept	n8n equivalent
Trigger	Trigger node (Webhook, Cron, app-specific)
Action	Action node
Filter	IF node
Path	Switch node
Formatter	Function / Set node
Delay	Wait node
Digest	Aggregate node
Sub-Zap	Sub-workflow (Execute Workflow node)

Example: Gmail → Slack + CRM

Zapier version (3 tasks per trigger):

1. Trigger: New email matching filter in Gmail
2. Action: Post message to Slack
3. Action: Create contact in HubSpot

n8n version (1 execution per trigger):

[Gmail Trigger] → [IF: matches filter]
                     → [Slack: Post message]
                     → [HubSpot: Upsert contact]

Same functionality. On Zapier, this is 3 tasks per trigger → 3,000/month at 1,000 runs. On n8n, it is 1 execution per trigger → 1,000/month. At n8n Pro ($50/mo for 10K executions), you have 10x the headroom.

Step 5 — Run in parallel for 7–14 days

This is the step teams skip and regret.

For each migrated zap:

1. Keep the Zapier version active
2. Turn on the n8n version too
3. Add a dedupe step in n8n so outputs don’t duplicate (tag records with source: n8n, filter on source: zapier in the destination to ignore)
4. Compare outputs daily for 7–14 days
5. Only deactivate the Zap after a full week of clean parity

For triggers you can’t easily dedupe (e.g., Slack messages), route n8n’s output to a private test channel first, then flip to production after parity is confirmed.

Step 6 — Handle the edge cases

Common sticking points:

Zapier-specific app connectors

Some Zapier connectors (rare but they exist) don’t have an n8n equivalent. Options:

• Use n8n’s HTTP Request node with the app’s API directly (5–30 min of effort)
• Use the Zapier webhook as a bridge (let Zapier catch the trigger, POST to n8n)
• Check the n8n Community Nodes registry — many niche integrations exist

OAuth re-authentication

Every integration needs re-auth in n8n. Budget 30 seconds per credential. For 30 integrations, that’s ~15 minutes of clicking — but plan for it.

Zapier Formatter steps

Zapier’s Formatter is powerful; n8n uses a Function node (JavaScript). Most Formatter operations become 1–3 lines of JS. For non-developers, n8n’s Set node covers the basic cases (string manipulation, date formatting) without code.

Error notifications

Zapier emails you when a zap fails. In n8n, use the Error Trigger workflow pattern:

1. Create one workflow with an Error Trigger
2. Have it post to Slack or email on any failure
3. Reference it in your workflows’ settings as the error workflow

This is actually better than Zapier’s default — centralized error handling for every workflow.

Step 7 — Optimize post-migration

After switchover, n8n lets you do things Zapier couldn’t:

1. Batch operations. n8n’s Split In Batches node processes thousands of items with no per-item cost penalty. Huge for data syncs.
2. AI agent nodes. Replace “if X then Y” logic with a real agent where it makes sense (e.g., support ticket triage).
3. Sub-workflows. Extract shared logic once; call it from many workflows. Saves duplication and executions.
4. Schedule complex retries. n8n’s Wait + retry logic handles rate-limited APIs without consuming executions.
5. Self-hosted LLM. Point the AI Agent node at a local Ollama endpoint. Zero API cost for classification and routing work.

Typical results (April 2026 migrations we’ve seen)

Team size	Zaps	Zapier cost before	n8n cost after	Savings
3-person startup	12	$49/mo	$20/mo (Cloud Starter)	59%
20-person SaaS	35	$399/mo	$40/mo (self-hosted)	90%
100-person agency	120	$1,450/mo	$150/mo (self-hosted k8s)	90%
Enterprise (200+)	300	$4,200/mo	$400/mo (self-hosted HA)	91%

Savings compound over time as workflows grow. Zapier scales linearly with task count; n8n self-hosted is flat until you hit real server limits.

What not to do

• Don’t hard-cut on day one. You will miss edge cases. Run parallel for 7–14 days.
• Don’t migrate the long tail first. Start with the zaps that justify the effort.
• Don’t skip credentials. OAuth re-auth seems boring; skip it and you’ll find out at 3am.
• Don’t underestimate AI rebuilds. If you’re converting Zapier AI steps to n8n agent nodes, budget extra time — the mental model is different.
• Don’t forget monitoring. Set up error notifications on day one. Silent n8n failures are the #1 post-migration complaint.

Final checklist

• Inventory and prioritize zaps
• Set up n8n (cloud or self-hosted)
• Migrate top 20% first
• Run parallel for 7–14 days
• Set up error-trigger workflow
• Re-auth all credentials
• Cancel Zapier (only after 2+ weeks clean parity)
• Celebrate: you just cut your automation bill 60–90%

Bottom line

Migrating from Zapier to n8n in 2026 is mostly a project-management exercise, not a technical one. The tooling has improved to the point where 90% of zaps have a clean n8n equivalent. The cost savings are real and compound over time. And once you’ve done it, you own your automation layer — which is the biggest win of all.