Automation Cost Calculator: Zapier vs n8n vs Make

This page contains one affiliate link, marked where it appears. The calculator itself is free, needs no signup, and does not send your numbers anywhere. Everything runs in your browser.

Every automation tool advertises a monthly price, and none of those prices tell you what you will pay. The reason is that the three big options do not count the same thing. Zapier counts tasks, n8n counts workflow executions, Make counts credits. Feed the same workflow into all three and you get three numbers that are not even the same order of magnitude.

This calculator does that arithmetic for you. Describe one workflow, get the monthly usage on each of the three, and see which plan it lands in. If you want the reasoning rather than the number, how the three platforms price at volume works through it in full.

Key Takeaways

• The three tools do not count the same thing. Zapier bills per successful action, n8n per workflow execution, Make per module action. One workflow produces three numbers that are not the same order of magnitude.
• Your bill scales with items processed, not with workflow length. A three-step workflow looping over 200 rows costs far more than a fifteen-step workflow that runs once and finishes.
• Six visible steps at 500 runs a month is 2,000 Zapier tasks, 500 n8n executions and 2,500 Make credits. Same work, three billing models, three different answers to which plan you need.
• Zapier does not bill triggers or its own built-in tools. Formatter, Paths, Filter, Delay and Looping are free, so any comparison that counts every visible step overstates what Zapier charges.
• This gives you usage, not your invoice. Usage is arithmetic and does not go stale. Prices do, so read the live pricing page before you buy.

The one field people get wrong

It is the second one, items per run.

Most people size their automation by counting the boxes on the canvas. That is the wrong unit. Your bill scales with how many items you process, not with how long your workflow is. A three-step workflow that loops over 200 spreadsheet rows will cost you far more than a fifteen-step workflow that runs once and finishes. If the number above came out higher than you expected, why a shorter workflow can cost more explains exactly which part of it is generating the charge.

This is where surprise bills come from, and it is well documented by the people it happens to. One Zapier user described a loop processing spreadsheet rows: “each step within the loop is consuming tasks individually. For example, if 5 rows are processed, the number of tasks used is significantly higher than expected.” A Zapier community manager confirmed in the same thread that the Looping, Filter and Delay actions themselves are not billed, and helped restructure the Zap down to two tasks per row. (Zapier Community, May 2025)

Set items per run to 1, then to 20, and watch what happens. Zapier and Make climb together. n8n does not move at all, because it charges for the run and not for what is inside it. That single behaviour is most of the reason people migrate to n8n at volume, and it is invisible if you only compare the headline prices.

How each number is calculated

The formulas are deliberately simple, and they are all here so you can check them.

Let R be runs per month, I items per run, A billable actions inside the loop, and B billable actions outside it.

What is not billable, and why it matters

Getting this right is the difference between a useful number and a scary one.

Zapier does not bill the trigger, and it does not bill its own built-in tools. Formatter, Paths, Filter, Delay, Looping, Sub-Zap, Digest, Zapier Manager, Storage, Tables and Forms are all free, and failed actions are not counted either.

This matters because the standard argument against Zapier counts every visible step in a Zap and arrives at a frightening number. That number is wrong. If you read a comparison claiming a six-step Zap costs six tasks per run, the author did not read the pricing page. The real gap on a typical freelancer workflow is closer to four to one against n8n rather than six to one, and four to one is still decisive. It is also defensible line by line, which the inflated version is not.

n8n bills per workflow execution. Node count does not enter into it.

Make bills one credit per module action. Its pricing page now says credits where the community and most articles still say operations. They are the same thing. Routers and error handlers are free, and filters sit on connections rather than being modules, so none of those count.

A worked example you can check by hand

A form submission arrives. A filter drops the incomplete ones. Then four things happen: create a CRM record, draft an invoice, send an email, append a row to a sheet. Six visible steps. It fires 500 times a month, one submission at a time. The 500 is an illustrative input; everything derived from it below applies each vendor’s own billing rules, retrieved from their pricing pages on 3 August 2026.

Now watch what that does to the plans. 2,000 tasks puts you past Zapier’s entry tier. 500 executions sits comfortably inside n8n’s Starter plan. 2,500 credits fits inside Make’s entry paid plan. Three tools, one workflow, and the cheapest option depends entirely on a billing rule that none of the marketing pages lead with.

Not sure what to automate yet?

Costing a workflow is the second decision. The first is whether the job is worth automating at all, and a surprising number are not. If you are still at that stage, read how to automate the manual parts of client work with AI first. It covers which jobs are genuinely worth handing over, the review step that keeps the output yours, and the two things that broke in my own workflow.

What the calculator deliberately does not do

It does not predict your invoice, and any tool that claims to is guessing.

It gives you usage, which is arithmetic and does not go stale, and it tells you which plan that usage lands in based on the tier limits published on the vendor pages, read on 3 August 2026 and re-checked where possible on 9 August. Prices and tier limits change, sometimes quietly. Above the entry paid tier all three vendors price by volume, so the calculator says to go and read the live page instead of inventing a figure.

It also assumes one workflow. If you run five, add up the usage from five passes.

Which one should you actually pick

Usage is one input to that decision and it is not the only one.

Pick Zapier if your automations are small and you never want to think about them again. It has the most guided builder, instant triggers as standard, and the largest app library. You are paying for not having to care.

Pick n8n if you run high-volume or fan-out heavy workflows and you are comfortable hosting something. Per-execution billing is the cheapest model that exists once you are past toy automations. The catch is real though: self-hosting on your own machine only runs while that machine is on, which quietly defeats the purpose for anything that has to fire at 2am. Budget around five dollars a month for a small VPS that is genuinely always on.

Pick Make if you want most of n8n’s power without the setup cost. It sits in the middle on both price and learning curve, and the human approval step is built in rather than something you assemble yourself. If you land here, Make’s free plan gives you 1,000 credits a month to test with, which is enough to find out whether the builder suits you before any money changes hands. (That is an affiliate link. It costs you nothing extra and we only use it because Make is the one of the three whose program we are actually in.)

For the full picture, including where the learning curve genuinely bites and what broke on a production workflow, read the longer comparison: n8n vs Zapier vs Make for freelancers.

Common questions