n8n vs Zapier vs Make: An Honest Comparison for Self-Hosting

Most comparisons of n8n, Zapier, and Make get the framing wrong. They line up integration counts, screenshot pricing pages, and declare a winner. That ignores the single variable that actually changes the math: whether you can run the automation engine on your own infrastructure.

Full disclosure — we build on n8n at Catalyst. We'll still tell you when Zapier or Make is the right answer, because for plenty of small businesses, it is.

The Pricing Models Aren't Comparable

Before picking a tool, understand that all three count usage differently. Zapier charges per task, where a task is any action in your workflow. If you trigger a workflow that checks a condition, enriches data, and updates your CRM, that counts as three tasks. Make charges per operation, which is similar but includes data transformations. n8n charges per workflow execution (cloud) or is unlimited if self-hosted.

That last difference matters more than any other number on a pricing page. A 10-step workflow running 10,000 times costs the same as a 2-step workflow in n8n. On Zapier or Make, every extra step multiplies your bill.

Real numbers as of mid-2026: Zapier Professional is $69/month for 2,000 tasks. You need the Team plan at $103.50/month for 2,000 tasks with multi-step zaps. At 50,000 tasks, you're looking at $448/month or more depending on your plan tier and overages. Make Pro is $16.67/month for 10,000 operations. The Teams plan at $34.17/month gives you 10,000 operations with advanced features. At 50,000 operations, you're at roughly $82-105/month. Self-hosted n8n at the same volume costs whatever your server costs — $0/month for unlimited operations. Your only cost is server hosting, typically $20-50/month on a VPS for moderate workloads.

The spread is real. One anonymized retrospective of a DTC ecommerce brand migrating 23 Make scenarios to self-hosted n8n saw tooling cost drop from $348/month on Make Teams to roughly $12/month on a Hetzner VPS, but credential and webhook recreation consumed about 40% of total project time. Self-hosting saves money. It doesn't save time on day one.

Self-Hosting Is the Only Question That Matters

Here's the cleanest decision filter. Can your business legally, technically, and operationally run its own automation server? If yes, n8n becomes hard to beat. If no, the choice collapses to Zapier or Make.

Why? Zapier is cloud only. Your data flows through Zapier's servers. No self-hosting option. Period. Make is cloud only for most users. They offer an on-premise version for enterprise customers, but it's expensive and requires negotiation. n8n offers full self-hosting as the default. Run it on your own servers, your own cloud infrastructure, or even a Raspberry Pi. Your data never leaves your network.

For anyone handling regulated data — HIPAA, financial records, EU customer data under GDPR — that single architectural difference can determine whether a tool is usable at all. Data sovereignty fundamentally separates the platforms. n8n offers self-hosting options with encrypted secret stores, LDAP integration, and advanced RBAC permissions — critical for regulated industries handling sensitive customer data. Zapier operates exclusively as a managed cloud service with SOC 2 compliance but no option for on-premise deployment, creating potential blockers for organizations with strict data residency requirements.

The practical floor for self-hosting is lower than people think. n8n's Community Edition is free – unlimited workflows, unlimited executions, all 400+ integrations. Managed n8n hosts like PikaPods or Elestio run $3-7/month and handle updates for you. A bare Hetzner VPS with Docker Compose is around $4.50/month if you're comfortable with a terminal.

But don't romanticize it. n8n recommends self-hosting for expert users. Mistakes can lead to data loss, security issues, and downtime. If you aren't experienced at managing servers, n8n recommends n8n Cloud. A self-hosted instance you forget to patch is worse than a Zapier subscription you grumble about.

Where Each Tool Actually Wins

Zapier wins on breadth and time-to-first-automation. Zapier leads with 7,000+ app integrations. If an app exists, Zapier probably connects to it. This is their strongest advantage and the main reason companies start with Zapier. If your business runs on a niche industry SaaS — some vertical CRM, a specialized scheduling tool, an obscure accounting app — Zapier is the only tool with a real chance of having a native connector. And the learning curve is genuinely flat. A non-technical operations lead can ship a working Zap in an afternoon.

Make wins on visual clarity at a fair price. Make has the best visual builder. Workflows are displayed as flowcharts with clear branching, loops, and error-handling paths. You can see exactly how data flows through your automation. Make.com offers 2,000+ integrations, which covers the vast majority of business tools. The gaps are typically niche industry apps. Make's HTTP module lets you connect to any API with a few clicks, partially closing this gap. If you have moderately complex workflows, a small ops team, no developers, and an aversion to per-task pricing — Make is the right call.

n8n wins on cost at scale, control, and AI workflows. n8n offers over 400 core nodes, 600+ community nodes, and unlimited custom API integrations in 2026. Unlike other automation tools like Zapier or Make, n8n prioritizes flexibility and control, allowing users to create tailored workflows for any API without hidden costs or restrictions. The integration count looks smaller on paper, but the HTTP and code nodes mean anything with an API is reachable. For AI work specifically, n8n has pulled ahead — n8n supports LangChain, self-hosted LLMs, and retrieval-augmented generation (RAG) setups. So if your team is dabbling in multi-agent workflows or building internal AI systems with sandboxed environments and model orchestration, this is your playground.

Where Each Tool Quietly Fails

Zapier's failure mode is the bill. The pricing scales linearly with workflow complexity, so the moment your automations grow beyond simple two-step triggers, costs balloon. It also offers genuinely limited code capabilities — Zapier's "Code by Zapier" module supports JavaScript and Python but with significant restrictions including 6MB input/output limits, inability to install external packages, and limited execution time. If you need real logic, you'll hit a wall.

Make's failure mode is operations math. "Operations" billing trips up new users. Each module action including triggers and filters counts as an operation, so a 5-step workflow uses 5 operations per run, not one. Polling triggers can drain operations fast. A scenario polling every minute uses around 43,000 operations a month before doing any actual work. Teams routinely blow through plan limits without realizing it.

n8n's failure mode is the people problem. The UI is sleeker than Make's and looks deceptively simple — until you realize it expects you to know what you're doing. It tosses you into variables and expressions pretty quickly, which can be powerful if you're a developer, or terrifying if you just want to automate a lead handoff. If you've got a solid dev team and want full control, n8n can deliver. But you'll need someone who can manage servers and write JavaScript in their sleep. Self-hosting also means you own the uptime. If your VPS goes down at 2 a.m., nobody at n8n is paging an engineer for you.

The Honest Decision Framework

Forget feature matrices. Answer three questions in order:

Do you have access to a developer or technically confident operator? No → Zapier or Make. Yes → continue.

Do you have regulatory, data residency, or cost-at-scale reasons to self-host? Yes → n8n self-hosted. No → continue.

Will your workflows routinely exceed five steps or rely on heavy branching? Yes → n8n cloud or Make. No → Zapier is fine, and you'll ship faster.

Most small businesses start somewhere and migrate later. That's a reasonable strategy. Zapier on day one, n8n by the time you're running ten production workflows. The migration cost is real, but it's bounded — and once you're on a self-hosted execution-based model, your automation budget stops growing with your automation footprint.

If you want help mapping out which tool fits your operation — and which workflows are worth automating in the first place — see how we approach automation projects.