Guide · Automation tools · Melbourne, October 2026
Somewhere in your business there’s an automation called “Copy of Copy of New Zap (2)”. Greg built it in 2023. Greg has since left. It runs a few thousand times a month, and everyone walks past it the way you’d walk past a sleeping dog. This guide is the honest version of custom automation vs Zapier, Make and n8n: what each one is good at, how each one bills you, the signs you’ve hit the no-code ceiling, and when the right answer is “leave it alone”.
Fixed price A$1,950 + GST. We’ll tell you if your Zaps are fine. Often they are.
Short answer: Custom automation vs Zapier isn’t a fight you have to pick a side in. Zapier, Make and n8n are the right tools for simple, low-risk workflows, and most businesses should start there. Custom automation earns its keep when a workflow moves money or personal information, has lots of exceptions, runs at a volume where per-step pricing bites, or has quietly become something only one person understands. The best answer is usually a mix: no-code for the glue, custom for the core.
Side by side
If you only read one thing, read this table. It’s the whole argument, minus Greg.
| Zapier | Make | n8n | Custom-built | |
|---|---|---|---|---|
| What it is | Hosted no-code automation with the biggest app catalogue | Hosted visual scenario builder | Workflow tool you can use in their cloud or host yourself | Code written for your workflow, running on infrastructure you control |
| Who builds it | Anyone who can follow a wizard | A confident tinkerer | A tinkerer who isn’t scared of JSON | A developer, ideally one who writes things down |
| You pay per | Task (each action step that runs) | Credit (each module that runs) | Execution (one whole run, however many steps) | Build once, then hosting and upkeep |
| Paid plans from | Professional, US$19.99 a month billed annually, 750 tasks | Make plan, US$9 a month for 5,000 credits | Cloud Starter, €20 a month billed annually, 2,500 executions; self-hosted Community Edition is free | A one-off build, quoted per workflow |
| Complex logic and exceptions | Paths and filters; gets crowded fast | Routers, iterators and error handlers | Branches, loops, code nodes, error workflows | Anything, including tests for it |
| Where your data goes | United States (AWS); no EU or Australian option | US or EU (AWS), chosen at sign-up and fixed after | Cloud: Frankfurt, Germany; anywhere you like if self-hosted | Wherever you decide, including Australia |
| Best for | Simple glue between popular apps, fast | Multi-step workflows on a budget | Technical teams, high volume, data control | Workflows that carry money, personal information or most of your volume |
Prices are the first paid tier, with the billing period and currency as each vendor lists them, checked 10 October 2026. Full details and sources in the billing section.
Meet the contestants
Let’s get the confession out of the way. [Clears throat.] We like Zapier. We like Make. We like n8n. We’d say this even if nobody was paying us to build custom automation, which, to be clear, is our whole business model. These are good no-code automation products, made by people who solved a real problem: getting app A to talk to app B without hiring a developer.
The comparison people search for, Zapier vs Make vs n8n, usually ends in a feature table. That’s useful, but it misses the more important question: what kind of workflow are you trying to run, and what happens on the day it breaks? Each tool answers that differently.
The friendly one
The one your office manager set up on a Tuesday afternoon, and it worked first time.
The visual one
The one where someone says “look how pretty my scenario is” and turns their laptop around.
The technical one
The one your developer friend recommends, then offers to “just spin up a server for you”.
One more contestant: if your business runs on Microsoft 365, Power Automate may already be in a subscription you pay for. For simple Outlook, Excel and SharePoint workflows, check it before you buy anything. Our workflow automation page has the same advice, because it’s still true.
The meter
This is the bit most comparisons skip, and it’s the bit that turns a A$30 subscription into a “why is this A$900?” conversation. The three tools don’t just charge different prices. They count different things.
That’s why Zapier vs Make vs n8n price comparisons rarely agree: the same workflow can cost three very different amounts. Try it with one of yours.
Simplified: assumes every step runs on every run and each step uses one unit. AI steps, loops and branches change the count, and Zapier’s AI steps can use 5 tasks each. Units only, not a price: allowances and tiers are in the table below.
| Tool | Free plan | Entry paid plan | Next tier up | Worth knowing |
|---|---|---|---|---|
| Zapier | 100 tasks a month, two-step Zaps only | Professional: US$19.99 a month billed annually (US$29.99 monthly), 750 tasks | Team: US$69 a month billed annually (US$103.50 monthly), 2,000 tasks, up to 25 users | Checks for new data every 15 minutes on Free, 2 on Professional, 1 on Team. Over your limit, you switch to pay-per-task billing |
| Make | Up to 1,000 credits a month | Make plan: US$9 a month for 5,000 credits, more on a sliding scale | Enterprise: custom pricing | Runs as often as every 15 minutes on Free, every minute on paid. A run can last 5 minutes on Free, 40 on paid |
| n8n Cloud | Free trial only (no card needed); self-hosted is free, see below | Starter: €20 a month billed annually, 2,500 executions | Pro: €50 a month billed annually, 10,000 executions | Business (€667 a month billed annually, 40,000 executions) is self-hosted only |
List prices from each vendor’s pricing page, ex tax, smallest allowance at each tier. Prices change and allowances scale up in steps, so check the vendor’s page before you budget. Sources: Zapier, Make, n8n.
[Puts down coffee.] Here’s the thing nobody budgets for. The subscription is rarely the expensive part. The expensive part is the hour on a Thursday when the invoices workflow stops because a supplier changed their email template, and the only person who can fix it is in a meeting. Then the next Thursday. Then the one after that.
We cover that maintenance cost properly in the three-year cost section. For now, the takeaway: a long workflow that runs often is where per-step billing hurts, and that’s exactly the workflow that most needs to be reliable.
The ceiling test
Tick the ones that sound like your business. Be honest. Greg isn’t watching.
A rule of thumb from Lumeio’s own work, not a scientific instrument. Signs 2, 4 and 7 count for more than the others: any one of them on a workflow that matters is worth a proper look.
The other option
“Custom” sounds expensive and slow, like a building extension that’s somehow still not finished. It doesn’t have to be. Custom automation means the workflow is written for your process, in code, and runs somewhere you control. In practice that’s usually a small, focused service that does one job properly.
What you get that no-code automation struggles to give you:
What custom costs you is time up front and a relationship with whoever maintains it. A no-code workflow can be running by lunchtime. A custom one takes days or weeks, depending on how many systems it touches. And if the person who built it doesn’t write documentation, you’ve just swapped Greg for a more expensive Greg.
Custom automation vs Zapier isn’t really “code vs no-code”. It’s “who owns this when it breaks, and how would we know?”
In our experience the sensible setup is rarely all one thing. Keep no-code automation for the dozens of small jobs it does well: new form entries into the CRM, Slack or Teams notifications, a row in a spreadsheet. Build the one or two workflows that carry money, personal information or most of your volume properly. Self-hosted n8n sits nicely in the middle for technical teams: a visual tool, on your own infrastructure, with code where you need it.
The real numbers
Here’s an illustrative comparison for one busy workflow. It’s made up, but if it feels familiar, you’re probably not alone.
The workflow: supplier invoices arrive by email. The automation reads each one, looks up the supplier, checks the purchase order, creates the bill in Xero and posts a message to the accounts team. That’s 6 action steps, about 1,500 invoices a month. In Zapier terms, roughly 9,000 tasks a month.
| Cost | No-code (Zapier or Make) | Custom-built |
|---|---|---|
| Build | A few days of someone’s time, often unpaid and unrecorded | A one-off build, quoted per workflow |
| Subscription or hosting | Zapier Team at about 10,000 tasks a month: a higher task tier than the US$69 entry price, paid every month, rising as volume grows | Cloud hosting, often tens of dollars a month for a workflow this size |
| Fixing it when it breaks | 2 hours a month × A$55 = A$3,960 over 3 years | Included in a support arrangement, or about 1 hour a quarter |
| Errors that get through | Unknown, because nobody’s counting | Logged, flagged and visible |
| If the builder leaves | Greg’s in Portugal | Documented, version-controlled, handed over |
Illustrative only. A$55 an hour is wages plus on-costs, the same figure we use in our ROI of AI automation guide. Your numbers depend on your volumes, systems and how often things change. Indicative, not a quote.
Two things usually fall out of this table. First, at low volume no-code automation wins on cost, clearly. If the same workflow ran 100 times a month, we’d tell you to keep it on Zapier and spend the money on lunch. Second, the subscription line is rarely the deciding factor. The rows that decide it are “fixing it when it breaks” and “errors that get through”, and those are the ones nobody measures.
Our rule of thumb: a custom build should pay for itself within about a year from the hours and errors it removes. If it can’t, it’s the wrong project. For what builds and consultants actually charge in Melbourne, see our guide to AI consultant costs, which has our own prices on the page.
2 hours a month firefighting
Exceptions only
The AI question
All three tools now offer AI steps. Zapier has Agents, Copilot and AI by Zapier steps. Make has AI modules and an AI agent. n8n has AI agent and LLM nodes, which you can point at the model provider of your choice. It shows: the ABS reported that around 12% of Australian businesses used AI in 2024-25, rising to 22% of medium businesses. You can drop a “summarise this email” or “extract the invoice total” step into a workflow in minutes, and for low-stakes jobs that’s genuinely handy.
The catch is the same one we wrote about in our document AI vs OCR guide. An AI step that reads an invoice will sometimes be confidently wrong. A custom build can check every value against rules (does the GST add up? is this ABN a real supplier?), score its confidence, and send the doubtful ones to a person with the source document beside them. You can bolt some of that onto a no-code workflow, but by then you’ve built most of a custom system inside a tool that charges per step.
Our view: use AI for the parts that are genuinely ambiguous, like classifying, extracting and summarising. Use plain rules for anything predictable. And keep a person on anything that matters. That split doesn’t change whichever tool you choose. If the real question is whether to buy an AI product or build your own, our build vs buy AI guide has a scorecard for exactly that.
Human review
Automation, no-code or custom, should do the moving and checking. People should do the deciding. Here’s how we split a typical workflow.
None of this says you can’t use Zapier, Make or n8n. It means knowing which workflows touch personal information, where that data is processed, and who can see the connection settings. Read more about keeping people in charge in our explainer on human in the loop.
Common mistakes
We’ve made a couple of these ourselves. [Looks at shoes.] Here are the ones that cost the most.
Our take
We don’t sell Zapier, Make or n8n, and we don’t have a favourite. We start with how the work actually moves through your business, then pick the cheapest option that fixes it properly. Sometimes that’s a better-named Zap. Sometimes it’s Make or self-hosted n8n. Sometimes it’s custom automation for the one workflow that carries the money, with the no-code tool kept for everything else.
What we always add, whatever the tool: written documentation, alerts that reach a named person, a clear split between rules, AI and people, and a handover so that nothing depends on one person’s memory. Especially not Greg’s.
See how we approach it on our workflow automation page, browse the automation use cases, or read how the same thinking works for invoice processing in Xero and inventory reorders.
Who wrote this
FAQ
Not automatically. Zapier is the better choice for simple, low-volume workflows between popular apps, and it can be running in an afternoon. Custom automation is better when a workflow moves money or personal information, has lots of exceptions, runs at high volume, or needs tests, logs and version control. Most businesses end up with both: no-code for the small jobs, custom for the one or two that matter most.
It depends on how many steps your workflow has. Zapier charges a task for each action step, Make charges a credit for each module including the trigger, and n8n Cloud charges one execution per run, however many steps. For short workflows the difference is small. For long workflows that run often, n8n’s per-run billing is usually far cheaper, and self-hosted n8n has no usage meter at all.
When the workflow carries money or personal information, when failures reach customers before you notice, when only one person understands it, or when you’re upgrading plans for volume rather than features. Any one of those on an important workflow is worth a look. If none apply, stay put. A well-named Zap that works is a perfectly good piece of business software.
The Community Edition of n8n is free to self-host under n8n’s Sustainable Use License, which allows use for your own internal business purposes. You still pay for the server and for someone to update, secure and back it up. Features such as single sign-on, Git version control and multiple environments need a paid Business or Enterprise licence.
Zapier says it hosts data on AWS servers in the United States, with no EU or Australian option. Make offers US or EU, fixed at sign-up, and n8n Cloud stores data in Frankfurt. If a workflow handles personal information and your business is covered by the Privacy Act, APP 8 applies to sending it overseas. Self-hosting or custom code can keep data in Australia.
It depends on how many systems the workflow touches and how many exceptions it has, so it should be quoted per workflow after someone has looked at the process. A useful test is payback: a custom build should pay for itself within about a year from the hours and errors it removes. If it can’t, keep the no-code version.
For low-stakes jobs like summarising an email, often yes. For anything that feeds money or records, an AI step needs checks around it: rules that test the values, a confidence threshold and a person who reviews doubtful results. Note that as of October 2026 an AI by Zapier step can use up to 5 tasks, and the 5-task tier is the default for new steps.
A 30-minute fit call is free. The Pain Point Audit is A$1,950 + GST for a half-day walkthrough of your processes and existing automations, with a written report within 5 business days and a fixed price for the first build. The fee is credited in full against a first build of A$5,000 + GST or more signed within 60 days.
Bring us your automation account, all 37 of them. The Pain Point Audit is a half-day look at how work really moves through your business, with a written report in 5 business days: which workflows cost the most time, which no-code automations are fine as they are, which ones should be rebuilt, what stays with your people and a fixed price for the first fix. Greg is welcome to join remotely.
Fixed price A$1,950 + GST. Credited if you go ahead with a build.
