When to use n8n vs Make vs Zapier (a decision tree).

If you're picking an automation platform for an Australian SMB right now, you have three real options: n8n, Make (formerly Integromat), and Zapier. We've shipped workflows on all three across the last 18 months. Here's the decision tree we walk every client through.

Start with three questions.

Before picking a tool, agree on these three things with the business owner:

• Volume. How many runs per month, in the steady state? 100, 10,000, or 1,000,000?
• Complexity. Is this a linear A-to-B-to-C workflow, or does it have conditional branches, loops, error retries, custom code blocks?
• Ownership. Who maintains this after we hand it over? A non-technical owner, an internal ops manager, a developer-adjacent staff member, or no-one?

Zapier — when you should pick it.

Pick Zapier when the workflow is simple, the volume is low, the ownership is non-technical, and the business will live happily with a higher per-run cost in exchange for the simplest possible UI.

Zapier's strength is its UX and its sheer integration count. It has a connector for almost every SaaS tool an Australian SMB uses, often more polished than the alternatives. Its weakness is per-task pricing that gets expensive fast above a few thousand runs per month, and limited branching/looping in the standard plan.

Sweet spot: a five-step linear workflow, 100 to 2,000 runs per month, owned by a non-technical office manager who needs to edit it without help.

Make — when you should pick it.

Pick Make when the workflow has real branching logic, the volume is mid-to-high, the ownership is a slightly technical operations person, and the business cares about visibility into what each step is actually doing.

Make's scenario builder is the best at showing you, visually, what's flowing through a workflow at each step. Per-operation pricing is meaningfully cheaper than Zapier above a few thousand monthly tasks. The trade-off: the UI has more learning curve, and the integration library, while extensive, is sometimes shallower than Zapier's.

Sweet spot: a 15-step workflow with three conditional branches, 5,000 to 50,000 runs per month, owned by an internal ops manager who'll inspect runs in the dashboard when something looks off.

n8n — when you should pick it.

Pick n8n when the workflow is complex, the volume is high, custom code is required for at least one step, and either you have an internal developer or you're comfortable having Belver.AI maintain it.

n8n is the only one of the three you can self-host (or have us host for you). At scale, the cost difference is enormous: you're paying for compute, not per-task. It also has native JavaScript and Python execution nodes, which matters when an automation needs to do anything custom. The trade-off: it's the most developer-flavoured of the three, and a non-technical owner won't comfortably maintain it without help.

Sweet spot: a multi-branch workflow with custom data transformations, 50,000+ runs per month, an Australian SMB that's professionalised its ops function and is happy with us (or their own developer) owning the platform.

The pattern we recommend most often.

If you're a small business reading this and you don't have a strong opinion already, default to Make. It's the best balance of capability, cost and learnability for the bulk of SMBs we work with. Zapier is the right choice when the workflow is genuinely simple and ownership has to be non-technical. n8n is the right choice when scale, complexity or cost rule the others out.

Whichever you pick, write down which question (volume, complexity, ownership) drove the decision. When the business outgrows the choice 18 months later, you'll know which knob to revisit first.