If you've outgrown a SaaS automation tool and you're looking at switching to n8n, the question is usually some combination of pricing, control, and code escape hatches. This is a head-to-head between n8n and Make.com drawn from running both — not affiliate copy.
The short version
| Pricing | n8n: free self-hosted; n8n Cloud from $20/mo. Make: free 1,000 ops/mo; Core $9, Pro $16, Teams $29 — billed by 'operations' (one node execution = one op). |
| Hosting | n8n: self-hostable (Docker, $5 VPS) or n8n Cloud. Make: cloud-only. |
| Integrations | n8n: 400+ native + HTTP node for anything else. Make: 1,500+ native modules, generally deeper per-app. |
| Editor | n8n: graph-based; can be linear or DAG. Make: scenario-based with native bundle iteration; great visual UX for branching. |
| Code escape hatch | n8n: Code node (JS/Python), full Function node, Custom node SDK. Make: limited — small JSON tools and HTTP/webhook modules; no full code node. |
| Self-hosting | n8n: yes, primary mode of operation for many users. Make: no. |
When n8n is the right call
- You want to self-host. If running infra is in your wheelhouse, n8n on a small VPS gives you unlimited workflows for the cost of one box.
- You hit task / op limits in Make.com. Pricing on usage-billed automation grows fast. Self-hosted n8n is a fixed cost.
- You need a real code node. n8n's Code/Function nodes give you full JavaScript and Python — useful when the visual editor isn't enough.
- You need AI agents. n8n's LangChain integration is more mature than Make.com's equivalent for chained / agent patterns.
- You want JSON-portable workflows. Every n8n workflow is a JSON file you can version-control, share, and import into any other n8n instance.
When Make.com is still the right call
- You don't want to run infra at all. Hosted SaaS with no ops is a real benefit when your team is non-technical.
- You need an integration n8n doesn't have native. Use the HTTP node — but if you're going to write a lot of those anyway, Make.com may still be faster to get started.
- You're a single-person workflow operator who lives in Make.com already. Migration cost matters; sometimes the ergonomics you know win.
How to migrate (if you're switching to n8n)
- Spin up n8n. Run
docker run -p 5678:14,000+ n8nio/n8non a Hetzner / Fly / Railway box, or sign up for n8n Cloud at $20/mo. - Pick the most painful Make.com zap to migrate first. Usually the one hitting task limits or the one with branching that Make.com doesn't handle cleanly.
- Find a starter workflow on AutomationFlows. The catalog has 14,000+ free workflows tagged by integration. Search the integrations involved in your existing zap.
- Copy the JSON, import into n8n, swap your credentials. AutomationFlows strips credential IDs before publishing, so you'll add yours fresh.
- Run both in parallel for a week. When you're confident, kill the Make.com side.
Useful starting points in the catalog
Once you're on n8n, here are the highest-leverage workflow categories to browse:
- AI & RAG workflows — agents, RAG pipelines, vector DB ingestion
- Email automation — triage, lemlist replies, transactional sends
- Slack & Telegram — bots, alerts, AI assistants
- E-commerce — Stripe, Shopify, order automation
- Data & Sheets — Google Sheets, Airtable, Notion, Postgres ETL
New to n8n? Skim the glossary first, then the import guide. The full methodology covers how the catalog itself is built.
FAQ
Is n8n really cheaper than Make.com?
For most teams above the free tier, yes. Make.com bills by tasks or operations, which compounds fast. Self-hosted n8n is a fixed cost (one VPS) regardless of throughput. n8n Cloud at $20/mo is competitive with their starter tiers.
Can I migrate my existing automations?
There's no automatic conversion — the data shapes don't map 1:1. The realistic path is rebuilding workflow-by-workflow, starting with whichever is currently most painful. AutomationFlows' 14,000+ free workflow templates often cover the same patterns.
Do I lose anything by switching to n8n?
You lose hosted-SaaS convenience (no ops). You gain ownership, fixed pricing, code escape hatches, JSON-portable workflows, and self-hostable AI integrations.