n8n vs Make vs Zapier: which automation platform fits (and when to go custom)?

n8n vs Make vs Zapier: which should you choose?

Choose Zapier for fast, simple automations across popular SaaS apps; Make for visual multi-step scenarios at lower cost per step; and n8n when you need self-hosting, onshore data or developer-level control. All three are good products. The wrong choice usually shows up six months in, as a bill that scales badly or a compliance question nobody can answer.

If your organisation runs on Microsoft 365, Microsoft Power Automate is also worth a look; we don’t compare it in detail here, so check its licensing and where your environment’s data is stored before deciding.

The table below reflects each vendor’s own pages as at 28 September 2026. Prices change often, so we describe the pricing model and link the live pricing pages in the sources rather than quoting dollar figures.

	Zapier	Make	n8n
What you pay for	Tasks: each successful action step. Triggers, filters, paths and the Formatter don’t count	Credits: each module action in a scenario counts as one	Executions: one full workflow run, however many steps it has
Plans	Free, Professional, Team, Enterprise	Free, Make plan, Enterprise	Cloud Starter, Pro, Enterprise; self-hosted Community (free), Business, Enterprise
Free tier	100 tasks a month, two-step Zaps only	1,000 credits a month	Self-hosted Community edition (you pay for hosting)
App catalogue	9,000+ apps	3,000+ apps	Smaller catalogue, plus generic HTTP and code nodes for anything with an API
Where it runs	AWS in the United States	AWS, US or EU data centre chosen per organisation	n8n Cloud on Microsoft Azure in the EU, or anywhere you self-host
Australian region	No	No	Yes, if you self-host in AWS, Azure or Google Cloud Australian regions
Self-hosting	No	No (an on-prem agent can reach internal systems)	Yes: Docker, Docker Compose, npm, or a cloud provider
Custom code	Code steps in JavaScript or Python	Custom JavaScript functions (plan-dependent; check the current plan table)	JavaScript and Python code nodes, custom nodes
Best at	Breadth, ease, non-technical teams	Complex branching at a moderate price	Control, data location, AI agents, high step counts

n8n vs Make vs Zapier pricing: how do the models change the real cost?

The counting unit matters more than the headline price. A workflow with many steps is cheap on n8n (one execution) and expensive on Zapier (one task per action). A workflow with one step that runs thousands of times a day costs about the same unit count everywhere, so plan price per unit decides it.

Take one illustrative workflow: a new order arrives, the customer is looked up in the CRM, the order is written to the accounting system, a Slack message goes to the warehouse, and a confirmation email is sent. That is one trigger and four actions.

Monthly volume	Zapier tasks (4 per run)	Make credits (about 5 per run, trigger included)	n8n executions (1 per run)
500 orders	2,000	about 2,500	500
5,000 orders	20,000	about 25,000	5,000
50,000 orders	200,000	about 250,000	50,000

Plug those unit counts into each vendor’s current pricing page to compare. Two things to watch. First, Make counts the polling trigger module too, and scheduled scenarios that check for new data every few minutes use credits even when nothing has changed. Second, error retries and loops multiply units on Zapier and Make, but not on n8n.

At low volume, the difference is small and ease of use should decide. At high volume, the per-step models can cost several times more than a per-execution model, which is when teams start looking at n8n or custom code. Our workflow automation cost guide covers the full cost, including the build and maintenance time these subscriptions don’t include.

Where does your data go, and does that matter in Australia?

At the time of writing, none of the three hosted services runs in Australia, so every record that passes through Zapier or Make, and n8n Cloud, is processed overseas. Zapier states it is hosted on AWS in the United States. Make lets you choose a US or EU data centre when you create an organisation, and you can’t change it later. n8n Cloud is hosted on Microsoft Azure in the European Union.

For many workflows that is fine. The Privacy Act 1988 doesn’t prohibit overseas processing. But APP 8 makes you responsible for taking reasonable steps so the overseas recipient handles personal information consistently with the Australian Privacy Principles, and you generally remain accountable for its mistakes. Automation platforms also keep execution logs, which may contain full copies of the data that passed through each step.

It matters more when:

• The data is health, financial or children’s information, or covered by a contract requiring onshore storage (common in government, health and financial services work).
• You’re an APRA-regulated entity and the platform becomes part of a critical operation.
• Your security review asks where logs are retained and who can read them.

Self-hosting n8n in AWS Sydney (ap-southeast-2), Azure Australia East or Google australia-southeast1 is the most direct way to keep automation data onshore without writing everything from scratch. Our data residency explainer covers the difference between keeping data onshore and keeping it out of foreign legal reach.

What does self-hosting n8n in Australia involve?

Self-hosting n8n in an Australian region gives you control over location and cost, and hands you every operational job the vendor used to do. n8n’s documentation offers Docker Compose for production, a one-line setup for quick starts, npm, and guides for AWS, Azure, Google Cloud and DigitalOcean.

What you take on:

1. A server and a database. A small container plus PostgreSQL for production, sized for your concurrency.
2. Patching. n8n releases frequently. Someone has to upgrade it and test workflows afterwards.
3. Backups and recovery. Workflows, credentials and execution history all need backing up, and restoring needs rehearsing.
4. Security. TLS, access control, secrets handling, and keeping the editor off the open internet.
5. Monitoring. Alerts when workflows fail or the queue backs up.
6. Licence choices. The free Community edition excludes SSO (SAML, LDAP), environments, external secrets, log streaming, projects, workflow and credential sharing, and Git version control. Those need a Business or Enterprise licence. The Community edition is licensed under n8n’s Sustainable Use License, which permits internal business use but isn’t an OSI open source licence, so read it before building a product on top.

If nobody on your team is comfortable with those six items, n8n Cloud or Make will be cheaper in practice than a self-hosted instance that quietly stops working.

When is Zapier or Make the better choice?

For most small teams automating SaaS tools, a hosted platform is the right answer, and paying a developer would be a waste. Choose Zapier or Make if:

• The workflows connect mainstream SaaS apps (CRM, email, forms, accounting, Slack) and move modest volumes.
• A non-technical person needs to build and change them without a developer.
• Nobody wants to run infrastructure.
• The data isn’t sensitive, or overseas processing has been assessed and accepted.

Zapier edges it for breadth and ease; with 9,000+ apps, it probably has a connector for the obscure tool your team uses. Make is the better fit once scenarios branch, loop and transform data, and its visual editor makes complex flows easier to follow.

When should you go custom instead?

Write code when the workflow is part of your product, carries real volume or risk, or needs engineering discipline that no-code tools make hard. The signals:

Signal	Why it points to custom
Monthly platform bill grows faster than revenue	Unit pricing on per-step models scales linearly with volume; compute for a coded service barely moves
The workflow is customer-facing or revenue-critical	You need automated tests, staged releases and rollback
Several people edit workflows and things break	Code gets version control, review and a history of who changed what
Data must stay in Australia under a contract or regulation	Runs wherever you deploy it, with logging you control
Complex logic, long-running jobs or heavy data processing	Platforms impose timeouts and step limits; code doesn’t
AI steps that need evaluation and cost tracking	A coded service can log prompts, score outputs and cap spend

Common triggers for moving off a no-code platform are a monthly bill that keeps climbing with volume, workflows nobody can safely change, failures that go unnoticed, and a compliance review asking where data goes. Custom doesn’t have to mean all or nothing. A common pattern is to move the two or three critical, high-volume flows to a small coded service in your own cloud account and leave the long tail of internal automations on Zapier or Make, where non-technical staff can keep maintaining them. If you’re weighing AI steps, our comparison of AI agents, chatbots and workflow automation explains which problems need which tool.

A quick decision guide

1. Just starting, a handful of simple automations, no sensitive data: Zapier.
2. Multi-step scenarios with branching, cost-conscious, EU hosting acceptable: Make.
3. Data must stay onshore, or you have a technical team and high step counts: self-hosted n8n in an Australian region.
4. Want n8n’s model without running it: n8n Cloud, accepting EU hosting.
5. Workflow is core to the business, high volume, or needs testing and audit trails: custom code, possibly alongside one of the above.

How All Webbed Labs approaches automation

We start by listing every workflow, its volume, the data it touches and what happens when it fails. That inventory usually shows that most flows belong on an off-the-shelf platform and only a few justify code. Where they do, we build them as small services in your own cloud account, in an Australian region by default, with tests, monitoring and the source in your repository. We also deploy and harden self-hosted n8n when that’s the better fit. See our workflow automation service, or compare the broader trade-off in low-code vs custom development.