Most n8n vs Zapier vs Make comparisons list features side by side. A difference that gets less attention is how each tool counts usage, so the same workflow can produce very different bills. Zapier counts tasks, Make counts operations and bills them as credits, and n8n counts executions. Once you know how each one counts, the right tool for a given workflow is much easier to see.
This comparison is written from the builder’s side. It walks through one example workflow on all three platforms, then covers code, hosting and the situations where each tool fits best. Plans and prices change often, so check each vendor’s current pricing page before you commit.
How each platform counts usage
Zapier bills by task. Its help article on how task usage is measured defines a task as any successful action that runs in Zapier. Triggers never use tasks, and neither do Filter or Paths steps or Formatter steps. When an account reaches its task limit, Zapier either bills extra tasks under pay-per-task billing or holds new Zap runs until the billing cycle resets, depending on the plan.
Make bills by credits. Its help page on operations defines an operation as a module run to process or check data, and its page on credits says that for non-AI apps one operation equals one credit. Two details matter for builders. A trigger module counts one operation each time it checks for data. And every module after the trigger runs once per bundle, so if the trigger returns ten items, each later module runs ten times. Make’s own example is a scenario where ten form responses produce 31 operations in a single run.
n8n bills by execution. Its pricing page defines an execution as a single run of your entire workflow, and says the number of steps and the amount of data do not change that count. A two-step workflow and a twenty-step workflow each count as one execution per run.
One workflow, three counts
Take a common sales workflow. A new lead arrives from a web form. The workflow looks up the company, adds the lead to the CRM, posts a message to the sales channel and sends a welcome email. That is one trigger and four actions. Assume it runs 1,000 times in a month, with one lead per run.
On Zapier, the trigger is free and each of the four successful actions is a task, so each run uses four tasks and the month uses 4,000 tasks. If you add a Filter step to skip test leads, that step does not add tasks.
On Make, each run uses one operation for the trigger and one for each of the four modules, so five operations per run and at least 5,000 credits for the month. If the trigger polls on a schedule, it also counts an operation every time it checks for new data, including scheduled checks between leads, so the real number depends on how often the scenario is set to run. If you later change the workflow to process a list, for example ten line items per order, every module after the trigger multiplies by ten.
On n8n, each run is one execution, so the month uses 1,000 executions however many steps you add. Adding a sixth or seventh step does not change the count.
The pattern is what matters, not the exact figures. Zapier’s count grows with the number of action steps. Make’s count grows with steps and with the number of items flowing through them. n8n’s count grows only with the number of runs. Long workflows, and workflows that loop over many items, are where the counting models pull apart.
Writing code inside a workflow
All three let you drop into code when a built-in step is not enough, with different limits.
Curious how much an AI Agent could save your UAE business?
Calculate your exact Visa, Gratuity, and workspace savings compared to hiring a human.
Launch AI ROI CalculatorZapier’s code steps run small snippets of Python or JavaScript, according to its help article on using JavaScript in Zaps. Scripts have time and memory limits that depend on the plan, there is a 6 MB limit on input and output, and public npm packages can be added on paid plans.
Make’s Make Code app runs JavaScript and Python. It is available on paid plans, importing custom libraries is limited to Enterprise, and it costs two credits per second of code execution time.
n8n’s Code node supports JavaScript and Python. On n8n Cloud, Python code cannot import libraries. On a self-hosted instance, Python can import standard library and third-party modules if the runner image includes and allows them.
Hosting and who runs the server
n8n can run on your own infrastructure. Its hosting documentation says that without a licence key, n8n runs as the free Community edition, and that you can self-host on your own servers, on-premises or in a private cloud. The trade-off is in its guide to choosing how to use n8n: self-hosting needs technical expertise for installation and configuration, and you provide the infrastructure and handle maintenance yourself. For a team without someone comfortable installing and maintaining a server, n8n Cloud or one of the other two tools is the safer starting point.
If data location or full control over the environment is a hard requirement for your business, self-hosted n8n gives you that control. Check directly with Zapier and Make what hosting and data-location options your plan offers. If it is not a hard requirement, hosting should not be the deciding factor.
Which one to choose
Zapier fits teams that want many short workflows built by non-developers, where most Zaps have one or two actions and filters do a lot of the work. Its counting rewards that pattern, because triggers and filters are free.
Make fits builders who want a visual canvas with branching and data mapping, and whose scenarios handle a predictable number of items per run. Watch the bundle multiplication and the trigger checks, because they decide the bill more than the number of scenarios does.
n8n fits long workflows, high run counts, workflows with real code in them, and teams that want the option to self-host. Its execution count stays flat as a workflow grows.
Before you choose, count your own workflows the way each vendor does. List your main automations, the number of action steps in each, the items each run processes and how often each runs. A short list like that will tell you more than any feature table.
Moving from Zapier or Make to n8n
If you decide to move, start with the workflows that use the most tasks or credits, since that is where the counting difference pays off first. Our comparison of n8n and Zapier for UAE businesses looks at the same choice from a UAE angle, including WhatsApp integrations and where data is stored. If you would rather have someone look at your current setup first, Artin offers n8n and workflow automation services for businesses in the UAE and the US, and you can book a free strategy call.
Read next
If you are deciding where AI belongs in these workflows, our explainer on intelligent automation covers when plain automation is enough and when an AI step earns its place. If you are not sure automation will pay for itself at all, start with the AI ROI calculator and your own numbers.
