n8n vs Zapier vs Make for Australian Businesses in 2026

A disclosure before anything else: WebArt Design builds custom automation for a living. When a business outgrows these tools, we are one of the companies they call. So I have an obvious bias, and the honest thing to do is name it up front and then try to write the comparison I wish existed when clients ask me this question, because they ask it a lot.

Here is the other thing worth saying early. Most businesses asking “n8n or Zapier or Make?” should just pick one of the three and get moving. Custom automation is the right answer for a specific set of problems, and I will tell you exactly which ones at the end. For everything else, these platforms are good, they are cheap relative to developer time, and the worst decision is spending three months deciding.

The 30-second answer

Zapier if you are non-technical, your volumes are low, and you need the broadest app catalogue. Make if you want the cheapest hosted option at moderate volumes and someone on your team enjoys building things visually. n8n if you have any technical capability at all, care where your data lives, or expect your automation volume to grow, because its pricing model and self-hosting option age far better than the other two.

That is the short version. The rest of this article backs it up with real numbers in Australian dollars, and covers the one thing vendor comparisons always skip: where your data actually goes, and what the Privacy Act expects you to do about it.

The pricing models are the actual difference

Feature lists converge. All three platforms connect apps, run multi-step workflows, handle webhooks, and now bolt on AI steps. The durable difference is how each one charges you, because that determines what happens to your bill as you grow.

Zapier charges per task. Every action step in a workflow that runs successfully costs one task. The trigger is free, filters are free, but a five-step workflow burns four tasks every single time it fires. Zapier’s Professional plan starts at US$19.99 a month on annual billing for 750 tasks, which sounds fine until you do the arithmetic on a real workflow. A four-action Zap that fires 30 times a day is 3,600 tasks a month. You blow through the entry tier in six days.

Make charges per credit (renamed from “operations” in August 2025, so older articles you find will use the old word). One credit per module execution, and unlike Zapier, the trigger counts. A five-step scenario costs roughly six credits per run. Core starts at US$10.59 a month on annual billing for 10,000 credits, the cheapest hosted entry point in the category. The catch is the meter: iterators and loops burn one credit per record, so a scenario that loops over 50 rows costs 50 credits, not one. Extra credit packs also carry a 25% markup over your plan rate.

n8n charges per execution. One execution is one complete run of a workflow, no matter how many steps it contains. A two-step workflow and a forty-step workflow cost exactly the same. n8n Cloud Starter is €20 a month on annual billing for 2,500 executions, Pro is €50 for 10,000. And then there is the option the other two do not offer at all: the self-hosted Community Edition is free, with unlimited executions, on a server you control. You pay for the server (a small VPS runs AUD $10 to $25 a month) and you become responsible for keeping it running.

Note the currencies. Zapier and Make bill in US dollars, n8n Cloud bills in euros. None of them bills in Australian dollars, so every price on this page moves with the exchange rate. At the time of writing the Aussie dollar buys about US$0.69 and €0.61, and I have used those rates below.

What this costs at real volumes, in AUD

Abstract pricing tiers hide the story, so here is a worked example. Say you run a services business with a lead-handling workflow: a form submission comes in, the lead gets enriched, a CRM record is created, the sales owner gets notified, and a follow-up email goes out. Five steps. It fires 1,000 times a month.

Same workflow, same volume, a 7x spread between the cheapest and most expensive hosted option. Now double the lead volume, because that is presumably the goal. Zapier needs 8,000 tasks and pushes you up another tier. Make needs 12,000 credits and an extra pack. n8n Cloud is still inside the Starter allowance, and self-hosted n8n does not notice.

This is the structural problem with per-step pricing: your bill scales with how often the workflow runs, which usually means it scales with your success. The workflow did not get smarter at 8,000 tasks. It just ran more often.

To be fair to Zapier, you are paying for something real. Its app catalogue is the largest in the category at well over 7,000 integrations, its editor is the one a non-technical office manager can learn in an afternoon, and niche Australian SaaS tools often ship a Zapier integration first and sometimes only. If the one system your business depends on connects to Zapier and nothing else, the comparison is over and Zapier wins.

Who actually builds and maintains these things

Pricing assumes someone can build the workflows, and the three platforms make very different assumptions about who that someone is.

Zapier is built for non-developers. Linear workflows, plain-English configuration, the gentlest learning curve of the three. Its ceiling is real, though. Complex branching, data transformation, and error handling all get awkward, and you end up paying task overheads for steps that would be one line of code elsewhere.

Make sits in the middle. The visual canvas is powerful, and pleasant to work in once it clicks. Routers, iterators, and error handlers give it a depth Zapier cannot match. But “once it clicks” is doing some work in that sentence. Expect a week or two of frustration, and expect to be the person who understands the scenarios, because on the lower plans Make’s support is thin.

n8n is a developer tool with a visual layer. You can build a lot without code, but its best features assume you can write some: JavaScript or Python steps inside workflows, custom API calls to anything with an endpoint, and full control over error states. If nobody in your business can read code, n8n Cloud is workable but you will hit walls. If someone can, n8n does things the other two simply cannot.

One more n8n note, because clients keep asking if the company behind it will still be around in five years. It raised a Series C in October 2025 at a US$2.5 billion valuation, and Bloomberg reported in May 2026 that SAP took a stake valuing it at US$5.2 billion. Whatever else that means, the platform is not going anywhere.

Where your data lives and why the Privacy Act cares

Every one of these platforms sits in the middle of your systems, which means your data passes through infrastructure someone else controls. Where that infrastructure lives is not a footnote for an Australian business. It is a compliance question with a specific rule attached.

The Privacy Act sets out 13 Australian Privacy Principles, the APPs, which are the ground rules for how businesses must handle personal information. The eighth one, APP 8, deals with sending personal information overseas. Before you disclose personal information to an overseas recipient, you must take reasonable steps to ensure the recipient handles it in line with the APPs, and you generally remain accountable if they breach them. Liability follows the data. A lead’s name and email flowing through an automation platform’s offshore servers is exactly the kind of disclosure this rule is talking about.

Here is where each platform’s data actually lives.

Zapier hosts customer data, including the data processed on your behalf, on AWS servers in the United States. There is no EU option and no Australian option. Zapier’s own privacy documentation is upfront about this. Every Zap that touches personal information is a disclosure to a US recipient.

Make lets you choose an EU data zone when you create your account, so your scenarios and the data flowing through them run in European data centres. Still overseas from Australia’s point of view, still an APP 8 disclosure, but the GDPR is widely regarded as a substantially similar privacy regime, which makes the “reasonable steps” story easier to tell.

n8n Cloud runs on Microsoft Azure in Frankfurt, Germany. Same position as Make’s EU zone: overseas, but inside a strong privacy jurisdiction. Self-hosted n8n is the outlier in the whole category, because the data does not go anywhere. Run it on a Sydney AWS region or a server in your own rack and personal information never leaves Australian infrastructure. APP 8 does not even come into play for the automation layer.

Two reasons to care more about this in 2026 than you did in 2023. First, the tranche 1 privacy reforms are now in force, individuals have been able to sue directly for serious invasions of privacy since June 2025, and APP 11, the security principle, now expressly expects documented technical and organisational security measures. Second, the small business exemption is eroding. From 1 July 2026, an estimated 100,000-plus small businesses (real estate agents, lawyers, conveyancers, and accountants among them) came under the Privacy Act for the first time through the anti-money-laundering reforms, and the government has confirmed a second tranche of reforms is progressing that would remove the AUD $3 million turnover exemption more broadly. If your business has been ignoring APP 8 because you are under the threshold, that position has a shelf life.

None of this means you cannot use a US-hosted automation tool. Plenty of compliant Australian businesses do. It means the choice should be conscious: know what personal information your workflows carry, say so in your privacy policy, and prefer platforms that shrink the disclosure surface when the data is sensitive.

The decision table

When none of the three is the right answer

I said at the top that most businesses should just pick a platform. Here is the other side of that.

These tools are brokers. They hold OAuth tokens into your CRM, your accounting system, your email. Those permissions are usually far broader than the workflow needs, they accumulate, and nobody reviews them. When a workflow misfires and sends the wrong data to the wrong place, you are reading someone else’s error logs to find out what happened. For a business wiring up lead notifications, that trade-off is fine. For a business automating payroll data, health information, or financial transactions between core systems, it often is not.

There is also a capability ceiling. No-code platforms are brilliant at moving records between SaaS apps. They are much worse at anything with real logic in it: multi-system transactions that must succeed or fail together, processing that has to happen in a specific order at scale, integrations with legacy systems that have no modern API, or workflows where an audit trail is a legal requirement instead of a nice-to-have.

The progression we see with clients usually runs in steps instead of one big leap. A business starts on Zapier or Make, proves the value, then hits either the pricing curve or the capability ceiling. The next step is often a self-hosted n8n instance, professionally set up on Australian infrastructure with proper backups, monitoring, and access control, which keeps the visual tooling while fixing the cost and residency problems. I run a self-hosted n8n instance for our own internal workflows for exactly these reasons. And when the workflow becomes load-bearing, the final step is a custom automation build: purpose-written middleware that lives in your infrastructure, does exactly what your process needs, and produces logs you own. That is custom software development territory, and it costs more up front than any subscription, which is exactly why it is the last step and not the first.

Where I would start

If you are reading this because you have one or two workflows in mind and no automation running today: sign up for Make’s free tier or Zapier’s free tier this week, build the ugliest version of your workflow that works, and run it for a month. The lessons from a live workflow are worth more than any comparison article, including this one.

If you are already running automations and the monthly bill is climbing, count your steps and runs, then redo my worked-example maths with your own numbers. Businesses paying Zapier hundreds a month are often a weekend migration away from paying Make or n8n a tenth of that.

And if your workflows carry personal or financial data, or they have quietly become something your business cannot operate without, that is the point where the platform question becomes an architecture question. That is the conversation we have with clients most weeks, and it is worth having before the audit or the outage forces it.