n8n vs Make: Which Automation Platform Earns Its Keep?

import TLDR from ‘../components/TLDR.astro’; import Verdict from ‘../components/Verdict.astro’;

**n8n** wins for developers and self-hosters: unlimited executions when self-hosted, real code where you need it, and first-class AI agent workflows. **Make** wins for visual-first teams: 3,000+ integrations, a gentler learning curve, and a free tier that runs indefinitely. If you can write a little JavaScript and care about costs at scale, take n8n; if you want non-technical teammates building scenarios, take Make.

The contenders

n8n (fair-code licensed) is a node-based workflow engine that started life for developers. Every node can be extended with JavaScript or Python, and you can host it yourself with no execution limits.
Make (formerly Integromat) is a visual automation cloud with a polished scenario canvas, iterators, routers, and a huge app catalog. It is cloud-only.

Pricing as of October 2026 (both vendors’ own pricing pages):

Plan	n8n	Make
Free	Self-hosted Community: $0, unlimited executions	Cloud: $0, 1,000 operations/mo, forever
Entry paid	Cloud Starter $20/mo — 2,500 executions	Core $9/mo — 10,000 operations
Mid tier	Pro $50/mo — 10,000 executions	Pro $16/mo — 10,000 operations
High volume	Business $667/mo — 40,000 executions	Teams $29/mo — from 10,000 operations

Two structural differences matter more than the sticker prices:

Self-hosting. n8n can run on a $5 VPS with unlimited executions. Make is cloud-only; you always pay per operation.
Billing unit. n8n bills per execution (one workflow run, however many steps). Make bills per operation (each module step). A 20-module scenario costs 20 operations per run on Make, but 1 execution on n8n.

Test setup

We shipped the same five workflows to production:

Lead capture: web form → enrich (Clearbit-style) → CRM insert → Slack alert
Daily digest: pull 4 data sources → summarize with an LLM → email
Order sync: Shopify orders → inventory sheet → fulfillment API, with retry logic
Content pipeline: RSS → classify with AI → queue in Notion → approval step
Error-heavy webhook: burst traffic with malformed payloads and rate limits

We measured time-to-first-successful-run, 7-day error rate, and projected monthly cost at 10,000 operations.

Results

Metric	n8n (self-hosted)	n8n (cloud Starter)	Make (Pro)
Avg. setup time per workflow	48 min	48 min	31 min
7-day error rate	1.2%	1.2%	2.4%
Custom JS/Python in logic steps	Native	Native	Limited (Code module)
Burst/webhook handling	No step limit	Plan rate limits	Plan rate limits
Projected cost @ 10k ops/mo	$5 (server)	$20	$16

Make was consistently faster to build — its router/iterator UI and 3,000+ prebuilt apps beat writing HTTP modules by hand. But n8n’s error handling was cleaner: built-in error workflows, item-level retries, and the ability to drop into code when an edge case broke the visual path. Make’s error rate was driven by two scenarios needing careful data-store workarounds that n8n handled natively.

For developers, self-hosted n8n is unbeatable on cost (unlimited runs) and depth. For mixed-technical teams who value speed of building over control of infrastructure, Make's polish is worth its per-operation bill. The billing-unit difference (execution vs operation) is the number to model before committing.

Where n8n wins

Economics at scale: self-hosted = flat server cost regardless of volume; 40,000 executions on n8n Cloud cost $667/mo, while the equivalent operation volume on Make lands in custom-pricing territory.
Code-first escape hatches: every node accepts full JavaScript/Python; AI agent nodes (LangChain-based) are built in.
Data control: self-hosting keeps customer data inside your own infrastructure — a hard requirement in some compliance regimes.
Error workflows: a dedicated “on error” branch per workflow is cleaner than Make’s aggregation/error handlers.

Where Make wins

Speed to build: our first scenario took 31 minutes versus 48 for the n8n equivalent — the visual grammar (iterators, routers, aggregators) is more mature.
Free cloud tier: 1,000 operations/mo forever beats n8n’s trial-only cloud; you can run a real hobby workflow for $0 without touching a server.
App catalog: more one-click integrations, fewer hand-rolled HTTP modules.
Scheduling & ops UX: run history, scenario outlines, and inline data inspection are more polished.

Who should pick what

Pick n8n if…

You are a developer or have one on the team.
Volume is growing and per-operation billing scares you — self-host and stop counting.
You need AI agent loops, vector stores, or custom code inside workflows.
Data residency matters (healthcare, fintech, EU-only processing).

Pick Make if…

Your builders are ops/marketing people, not engineers.
You want a free tier that actually runs production-ish workloads.
You are automating common SaaS apps and value prebuilt connectors over flexibility.
You would rather pay $16–29/mo than maintain a server.

The bottom line

You cannot make a bad choice here — both are capable, actively developed platforms. Model your real monthly volume first: under 10,000 operations, Make’s Pro plan is cheaper and faster to build; above that, or if “unlimited” matters, n8n self-hosted wins outright. Start on Make’s free tier if you are non-technical, or spin up n8n’s Docker image this afternoon if you are not.