n8n vs Zapier vs Make in 2026: real-world picks for builders

"Should we use Zapier or n8n?" — asked weekly by every founder I talk to. The answer is almost always "it depends on three numbers". Let me show you which numbers.

The 30-second TL;DR

The three numbers that decide

Number 1: Runs per month

• 0–100/mo → Zapier free tier. Don't overthink. You'll spend more time evaluating tools than the time-saved is worth.
• 100–500/mo → Zapier Starter ($30/mo) OR Make Free (1000 ops free). Make wins if your workflows touch 3+ apps per run.
• 500–2000/mo → Make Core ($10.59/mo for 10k ops) is a steal. Zapier same volume = $73-198/mo. n8n self-hosted breaks even if you already pay for a $12 droplet.
• 2000+/mo → n8n self-hosted. The pricing math becomes silly otherwise.

Number 2: Workflow complexity

How many "if-then-else" branches and loops does your workflow need?

• 0-1 branches: Any tool works. Pick by price.
• 2-4 branches + light data transformation: Make or n8n. Zapier's Paths feature works but feels cramped.
• Custom code, loops, batched processing, error retries: n8n. Hands down. The JavaScript / Python nodes let you do real work.

Number 3: Data sensitivity

• Public data, marketing webhooks: Any tool fine.
• Customer data, internal docs, API keys: n8n self-hosted. Your droplet, your data plane, no third-party processing.
• Regulated (HIPAA, GDPR enterprise): n8n self-hosted with proper auth, or Zapier Enterprise ($$$).

Real-world picks I've made

MartialApps — instructor video upload notifications

3 runs/day max. Zapier free. Not worth the n8n setup.

Ease (mental health platform) — appointment reminders

~80 runs/day, customer health data. n8n self-hosted on the same droplet as the app. Zero per-task cost, data stays in our infra.

Personal — daily morning brief

1 run/day. Could have been Zapier. Used n8n because I already run it and wanted the JavaScript escape hatch.

Bodhua Bazar — order webhook → Slack + ClickUp

~20 runs/day in busy weeks. Make Free tier. Visual scenario, fan-out to 3 destinations, well under 1000 ops/mo.

The hidden cost of "free"

Zapier Free Tier has a real catch: only 100 tasks/mo and 15-minute polling for triggers. If your workflow needs to fire within 30 seconds of an event, you need a paid Zapier tier OR webhook-first tools (n8n, Make).

Make's free tier is more generous: 1000 ops/mo + true webhooks. For most early-stage projects this lasts longer than you'd expect.

n8n's "free" needs a server. Cheapest viable: $5/mo droplet (1 GB RAM). Tight but works for low-volume workflows. $12/mo gets you breathing room for 5-10 active workflows.

Switching costs

The longer you stay on a tool, the harder migration gets. Some pragmatic notes:

• Zapier → Make: One-click migration tool exists. Imperfect but a head start.
• Zapier → n8n: Manual rebuild. Plan a day per workflow.
• Make → n8n: Manual. Make's "scenarios" map cleanly to n8n workflows once you learn both UIs.
• n8n → cloud (Zapier/Make): Rare. Usually people move the other way.

My current recommendation matrix

Quick-decide flow:

Is the team non-technical AND below 100 runs/mo?
  → Zapier free.

Else: do you already pay for any cloud server?
  → If yes, install n8n. Marginal cost is zero.
  → If no:
      Below 1000 ops/mo? → Make free.
      Above? → Pick by complexity:
        Mostly linear → Make Core.
        Branched + code → n8n cloud ($20/mo) or self-host.

Where I land

For everything I build now, n8n self-hosted is the default. The droplet costs $12 anyway. I get unlimited runs, JavaScript escape hatches, my own data plane, and the workflows live in git as exported JSON. The Zapier-style "no-code" upside is gone, but I'm a developer — that was never the upside for me.

For clients without engineering, I still recommend Zapier or Make first. The maintenance burden of self-hosted n8n is real, even if the runtime cost isn't.

Want help choosing or migrating? Tell me your three numbers and I'll tell you which tool, with the math.