Zapier vs Make vs n8n: how I choose for client automations
Clients ask me this weekly, so here's the decision tree I actually use when scoping an automation — with the failure modes nobody mentions in the comparison blogs.
The 30-second answer
- Zapier — fastest to ship, most expensive per task, best when a non-technical team must own the workflow after I leave.
- Make — the visual-power middle: complex branching and cheap operations, still no-code enough for the team to read.
- n8n — maximum control, self-hostable, JavaScript everywhere; for teams with a technical person who'll maintain it.
How I actually decide
1. Count the tasks, then multiply by twelve
Zapier bills per task; a "simple" 5-step sync that runs on 2,000 records/month is 10,000 tasks before you blink. I model the client's real volume and price the platforms against a year of growth. Make's operation pricing usually wins this round by 3–10×; self-hosted n8n wins it by ∞ (server cost only).
2. Who maintains it in month six?
If the answer is "the ops person who likes spreadsheets" — Zapier's guardrails are worth the premium. If there's a developer on the team, n8n's code nodes and version-controllable workflows turn automations into real infrastructure. Make sits between: readable visuals, but no easy git story.
3. Where does the data live?
This one is non-negotiable for healthcare, finance, and legal clients. Zapier and Make are cloud-only. n8n self-hosted keeps every payload inside your own network — when a client's compliance sheet says "data may not leave our infrastructure", the decision makes itself.
The failure modes I've been bitten by
- Zapier: polling triggers with delay — for time-sensitive flows you need instant webhooks (fine, but plan for it) and the task bill compounds quietly.
- Make: scenario complexity explodes visually; a 40-module scenario is undebuggable by anyone except its author. Document as you build.
- n8n: you own the uptime. Self-hosted means updates, backups and monitoring are your job — I only recommend it with someone technical on the client side.
And when none of them fit
If the workflow needs custom parsing, retries with backoff, or talks to an API without a connector, a Google Apps Script system often beats all three: it lives where the client's data already lives (Sheets/Gmail/Drive), costs nothing at low volume, and there's no per-task meter at all. That's usually the moment a client hires me instead of a platform subscription.
Undecided which side your workflow lands on? Send me the steps you do manually — I'll tell you the platform, the monthly cost, and the honest answer if you don't need automation yet.