n8n vs Make vs Zapier. The same workflow costs different amounts, because each one counts something different.
Comparisons start from the plan price, which on its own tells you little: n8n counts workflow runs, Make counts credits, Zapier counts actions. Here is the decision table in Excel, with the maths for your own workflow.
For an SME, choosing between n8n, Make and Zapier depends on where the data has to stay, how often the workflow runs and who looks after it. n8n is the only one of the three you can install on a company server; Make lets you choose a European data zone; Zapier keeps data in the United States.
The counting rules are in the official price lists. n8n counts executions, meaning whole runs of the workflow; Zapier counts tasks, meaning successful actions; Make counts credits, one per operation of each module. A five-action workflow running a thousand times a month comes to 1,000 executions or 5,000 tasks.
Prices read on 13 November 2026 start at €20 a month for n8n, $9 for Make and $19.99 for Zapier, billed annually.
The decision table can be downloaded below, in Excel, with the maths for your workflow and a filled-in example, no email required.
Written by: Mattia Esposito, AI and marketing consultant, founder of Itria AI. From Puglia, based in Turin since 2023, he works remotely in Italy and abroad. LinkedIn profile.
How to build your first automation, from the trigger to the awkward cases, is in the guide on how to create an automation; how to write down the process before touching any tool, in the one on how to create a workflow. This page is about choosing the tool.
The n8n vs Make vs Zapier decision table, to download
An Excel file with four sheets: the comparison row by row with the source alongside, the maths for your workflow, eight questions to help you choose and a filled-in example. Prices come from the official pages, read on 13 November 2026, in the currency each provider shows: n8n in euros, Make and Zapier in dollars.

| File | Contents | Link |
|---|---|---|
| Excelto fill in, in Italian | The Comparison, My workflow, Questions and Example sheets. Monthly usage and the plan check calculate themselves. | |
| PDFto read, in Italian | The filled-in example: a five-action workflow, a thousand runs a month, counted in all three tools. |
| Sheet | What it does | What you fill in |
|---|---|---|
| Comparisontwelve rows | Unit counted, plans, currency, going over quota, users, data, self-hosting, licence, AI, announced changes. | The rows that change when you reread the price lists, with the date. |
| My workflowthe maths | Executions, credits and tasks per month; whether the plan is enough; the cost per thousand runs. | Runs per month, actions per run, modules in Make, plan quota and price. |
| Questionseight | Data, server, volume, agents, messages to customers, licence, connectors, upkeep. | Yes, no or not sure, with the consequence next to it. |
| Examplefilled in | An incoming order, five actions, a thousand runs a month. | Nothing: it shows how to read the maths. |
I taught myself automation, first artificial intelligence and then n8n: I say so on the about page. The transparency page also says where it runs: n8n installed on a Hetzner server in the European Union, on Itria’s own infrastructure, and not a third-party service.
So the table puts the three tools on the same rows, with the official page each cell comes from alongside it.
Where the automations run. n8n self-hosted on a Hetzner server in the European Union, under the provider’s data processing agreement. It is on the AI transparency page.
Who decides. “The machine prepares, a person always decides”: nothing goes out to a customer without approval, either of the content or of the single message. The rule is on the ethics page.
Who notices a stalled workflow. A system watches every automated cycle and raises an alert if a cycle does not finish within its expected window. This too is described on the transparency page.
n8n vs Make vs Zapier: the differences in one table
The differences that matter to an SME fit in eight rows: what gets counted, the first paid plan, the currency, going over quota, where the data sits, self-hosting, the licence and AI.
| Criterion | n8n | Make | Zapier |
|---|---|---|---|
| What gets counted | Executions: one whole run of the workflow. | Credits: one per operation of each module, more for AI. | Tasks: each successful action; the trigger does not count. |
| First paid plan | Starter, €20 a month billed annually, 2,500 executions. | Core, $9 a month billed annually, 10,000 credits. | Professional, from $19.99 a month billed annually. |
| Price list currency | Euros. | Dollars, even from Italy. | Dollars, with the euro among the selectable currencies. |
| Going over quota | Starter and Pro stop at the cap, with no overage charge. | To be checked on the official page. | Pay-per-task up to three times the plan limit, can be switched off. |
| Where the data sits | Cloud in Frankfurt; self-hosted, wherever you choose. | Data zone of your choice: US1, US2, EU1, EU2. | AWS servers in the United States. |
| Self-hosting | Yes, free Community Edition. | No. | No. |
| Licence | Sustainable Use License: free for internal use. | Subscription service. | Subscription service. |
| AI agents | AI Agent node and human approval on tool calls, from Starter. | Make AI Agent, in open beta since 2 February 2026. | Agents, with a separate activity quota. |
Anyone asking whether n8n is “just Zapier” is half right: both connect apps and services in workflows triggered by an event, such as an email or an order. What changes is the counting, where the data lives and whether you can keep the software on a server of your own.
n8n vs Make vs Zapier pricing units: executions, credits and tasks
A plan’s price only makes sense once you know what it counts. For n8n, an execution is one complete run of the workflow, however long. For Zapier, a task is each successful action. For Make, credits have replaced the old operations: for apps without AI, one operation equals one credit.
The example in Zapier’s help centre is the clearest: a Zap that uploads a file to Dropbox and posts a message in Slack uses 2 tasks per run, and ten runs make 20 tasks. The trigger is free, and so are tests in the editor, filters and paths. A run that stops at a filter therefore uses less than one that goes all the way.
With AI the sums get longer. AI by Zapier uses a different number of tasks depending on the model tier, and a successful call to Zapier MCP counts as 2 tasks. On Make, with your own connection to OpenAI, Claude or Gemini, you pay one credit per operation and the tokens go to the model provider; with Make’s AI Provider, credits depend on tokens.
From 1 December 2026, Next Gen Zaps and the SDK on Zapier use tasks too, one per external call, while Agents stay outside the task count with an activity quota of their own. If you are comparing now, redo the maths after that date.
n8n vs Make vs Zapier pricing in 2026
As of 13 November 2026, n8n starts at €20 a month, Make at $9 and Zapier at $19.99, all billed annually. Make and Zapier show their prices in dollars even from Italy; Zapier converts them to euros on request. Prices change, so check the official page before you decide.
| Plan | Price read on 13 November 2026 | What it includes |
|---|---|---|
| n8n Community Edition | Free, to install on your own server. | Internal use, under the Sustainable Use License. |
| n8n Starter | €20 a month billed annually, €24 billed monthly. | 2,500 executions a month, unlimited users and workflows. |
| n8n Pro | €50 a month billed annually, €60 billed monthly. | 10,000 executions a month. |
| n8n Business | €667 a month billed annually, €800 billed monthly. | 40,000 executions a month, self-hosted only. |
| Make Free | 0 $. | 1,000 credits a month, at least 15 minutes between runs. |
| Make Core | $9 a month billed annually, $9.54 billed monthly. | 10,000 credits a month, unlimited active scenarios. |
| Make Pro | $16 a month billed annually, $16.95 billed monthly. | 10,000 credits a month, priority execution. |
| Make Teams | $29 a month billed annually, $30.73 billed monthly. | 10,000 credits a month, teams and roles. |
| Zapier Free | 0 $. | 100 tasks a month, two-step Zaps, checks every 15 minutes. |
| Zapier Professional | From $19.99 a month billed annually, $29.99 billed monthly; in euros €18.28 and €27.42. | 750 tasks a month in the entry tier. |
| Zapier Team | From $69 a month billed annually, $103.50 billed monthly; in euros €63.09 and €94.63. | 2,000 tasks a month, 25 users, SAML single sign-on. |
Two details change the final bill. Paying month by month costs more: n8n states a 17% discount on annual billing, and Zapier labels it “Save 33%”. On Make, the prices in the table are for 10,000 credits a month, and rise with the credits. And going over quota works in opposite ways: on n8n’s Starter and Pro plans the workflow stops at the cap, while Zapier moves you to pay-per-task up to three times the limit, unless you switch it off.
Is n8n free? What it is and how it works
n8n connects apps and services in workflows triggered by an event, and runs either in the provider’s cloud or installed on a company server. The software is the same in both cases: without a licence key it runs as the free Community Edition, and a Business or Enterprise key unlocks the paid editions.
You install it with Docker Compose, on a server on your premises or on a private cloud such as AWS, Azure or Google Cloud. The software is free: the server, updates, backups and the hours of the person looking after it stay with the business.
The licence deserves a couple of lines, because comparisons often get it wrong. n8n uses the Sustainable Use License: you may use and modify it for your business’s own internal purposes, and paid consultancy building workflows for others is allowed. What you may not do without a commercial agreement is make n8n available to your own customers so they can connect their accounts to it.
This is why n8n does not call itself open source, but “fair-code”. The Open Source Initiative definition requires a licence not to restrict use of the program in any field of endeavour, and a licence that separates internal use from resale falls outside it. Until 17 March 2022 the licence was Apache 2.0 with the Commons Clause.
Self-hosted, something still leaves the server. With a Business or Enterprise licence, the key contacts n8n’s licence server every day with the number of executions, and telemetry is on by default, though it can be switched off.
Make vs Zapier: what they are and when they are enough
Make and Zapier are subscription services: no server to maintain, and the data sits wherever the provider keeps it. On Make a workflow is called a scenario and is made of modules; on Zapier it is called a Zap and is made of steps.
They are enough when the workflow is short, runs rarely, touches no data that must stay in Europe and nobody in the business wants to look after a server.
Between the two, the clearest documented difference concerns data: Make lets you choose a European zone, Zapier does not. The rest is decided on your own workflow, in the spreadsheet.
Where your data lives: GDPR and servers in Europe
The n8n cloud keeps data in the EU, in Frankfurt; Make lets you choose a European data zone; Zapier hosts data on AWS servers in the United States. Asked whether data can be kept only in the EU, Zapier’s privacy page replies: “Zapier does not support this option.”
| Where | What the provider says | What to check |
|---|---|---|
| n8n cloud | Data in the EU, on servers in Frankfurt, Germany. | The AI models connected to the workflow: each has its own provider. |
| n8n self-hosted | Wherever you put the server. | Telemetry and the licence key, both of which contact n8n. |
| Make, EU1 or EU2 zone | The organisation chooses its data zone. | Maia and web search go through OpenAI in the United States; some Claude models through the United Kingdom. |
| Make with your own AI connection | Make’s EU guarantee does not cover that call. | Your contract with the model provider. |
| Zapier | AWS servers in the United States; transfers under the Data Privacy Framework and standard contractual clauses. | Retention of 7 to 30 days only on the Company and Enterprise plans. |
Make puts it plainly in its help centre: “Selecting the EU1 or EU2 data zone does not automatically mean every request gets processed and stays in the EU.” For the latest Claude models served by Make’s AI Provider in the EU zone, on the other hand, processing stays in European regions, including Frankfurt, Zurich, Stockholm and Milan.
For Zapier, transfers to the United States rest on the EU-US Data Privacy Framework, the adequacy decision the European Commission adopted on 10 July 2023, plus a data processing agreement with standard contractual clauses. For a British business Zapier points to the UK Extension of the same framework; for an American one the question does not arise.
A server in Europe does not settle the matter on its own: the provider’s home country counts too. Why is explained in the guide on the CLOUD Act and artificial intelligence, and the questions to ask before pasting customer data into a service are in the one on ChatGPT, GDPR and the CLOUD Act.
n8n vs Make for AI agents: agents inside your automations
All three now put an AI agent inside workflows, but at different stages. n8n lists in its price list the AI Agent node, MCP servers, evaluations and human approval on tool calls, included from the Starter plan, and connects models from OpenAI, Anthropic and Google. Make AI Agent has been in open beta since 2 February 2026. Zapier has Agents, with a quota of their own.
The most honest advice comes from Make, about its own product: give the agent tasks “that you trust an intern to handle”, and keep it away from sensitive data, high-risk financial or strategic decisions and strict legal requirements. It holds for all three tools.
On 9 October I wrote on LinkedIn about 28 days thrown away with nine AI agents voting before every trade: tested on 1,212 trades, on average they made nothing. The turning point came when they stopped deciding and started researching and checking. I closed it like this: “AI cannot decide on its own. It researches, checks, remembers.” The full story is in the LinkedIn post (in Italian).
In a business workflow, this means a person approving before anything goes out to a customer or touches money: human in the loop. Two typical cases are in the guides on how to extract data from PDFs and invoices with AI and how to write quotes with artificial intelligence; designing an agent, in the one on how to build an AI agent.
How to choose between n8n, Make and Zapier
Task first, tool second: write the workflow down, count its runs and steps, decide where the data has to stay. With those answers the choice usually makes itself, and the My workflow sheet puts the numbers alongside.
If the data must stay in the EU or on a company server, look at n8n, in its Frankfurt cloud or self-hosted, and at Make in the EU zone, checking the AI features. If the workflow is long and runs thousands of times a month, n8n counts the whole run; for the other two, do the sums in the spreadsheet. If nobody in the business maintains a server, the cloud is what remains.
To try a two-step connection, Zapier’s free plan is enough. If you want to offer the platform to your own customers, n8n needs a commercial agreement. If the program you need to connect has no ready-made connector, the route may be a different one: the guide on RPA or an AI agent explains the difference.
On 5 October, writing on LinkedIn about the seven months in which I built Itria, I gave the advice I had not followed myself, after three months spent on the first product before showing it to anyone: “The solution has to fit THAT problem. Don’t over-engineer.” The post (in Italian) is about businesses, but it holds for workflows too: the most powerful tool is wasted on a two-step job.
What a workflow is and which parts it has is in the glossary, under workflow. The other guides on marketing and AI for small businesses are gathered on the marketing for SMEs page.
Questions and answers
n8n, Make or Zapier: which is better for an SME?
It depends on where the data has to stay, how often the workflow runs and who looks after it. n8n is the only one of the three you can install on your own server, and its cloud keeps data in Frankfurt. Make lets you choose a European data zone. Zapier hosts data in the United States.
On long workflows that run often, the way each one counts matters: n8n counts the whole run, Make and Zapier count every step.
Is n8n free?
The Community Edition is: you install it on your own server and, without a licence key, it runs free of charge. Its licence, the Sustainable Use License, allows use for the business’s own internal purposes. The n8n cloud is paid, with a trial of 1,000 executions and no credit card.
How much does n8n cost per month?
As of 13 November 2026, billed annually: Starter at €20 a month for 2,500 executions, Pro at €50 for 10,000, Business at €667 for 40,000, self-hosted only. Billed monthly: €24, €60 and €800. Prices change, so check the official page.
Is n8n just Zapier?
They do the same job, connecting apps and services in workflows triggered by an event, but they differ in three ways. n8n counts workflow runs, Zapier counts successful actions. The n8n cloud keeps data in Frankfurt, Zapier in the United States. And n8n can also be installed on your own server, under the Sustainable Use License, which n8n itself does not call open source.
Make or Zapier: which should you choose?
Between the two, the clearest documented difference concerns data: Make lets you choose a European data zone, EU1 or EU2, with some exceptions for AI features; Zapier hosts in the United States and says it does not offer the option of keeping data only in the EU.
As of 13 November 2026, billed annually, Make starts at $9 a month for 10,000 credits and Zapier at $19.99 for 750 tasks.
Notes on sources
- n8n, Plans and Pricing: plans, prices in euros billed annually and monthly, the 17% discount, the definition of an execution, data in Frankfurt, AI features, licence key and telemetry. Read on 13 November 2026.
- n8n, Sustainable Use License (the licence text) and Community license (licence FAQ, fair-code, consultancy, the licence before 17 March 2022).
- n8n Docs, Host n8n (Community Edition without a licence key) and Integrate AI (models from OpenAI, Anthropic and Google).
- Zapier, Plans & Pricing, annual and monthly prices in dollars and euros, read on 13 November 2026; Help Center, How is task usage measured in Zapier?, updated 13 November 2026: definition of a task, the Dropbox and Slack example, the change on 1 December 2026.
- Zapier, Data Privacy Overview: AWS servers in the United States, the quoted sentence, the Data Privacy Framework, data retention.
- Make Help Center, Credits, updated 13 November 2026; AI providers and data zones in Make, updated 11 November 2026, source of the quoted sentence; Introduction to Make AI Agent, updated 29 October 2026. Make, Pricing: prices in dollars for 10,000 credits a month, annual and monthly, read on 13 November 2026; there is no Italian version of the page.
- European Commission, EU-US data transfers: the adequacy decision on the Data Privacy Framework. Open Source Initiative, The Open Source Definition, criterion 6. Read on 13 November 2026.
- The facts about Itria come from the public pages about, AI transparency and ethics; the quoted sentences from the LinkedIn posts of 5 and 9 October 2026, in Italian.
You can pick a tool in an afternoon. A workflow that holds up for years is another matter. The system first, then the growth.
Mattia Esposito, who wrote this guide, builds tailored AI systems for small businesses with Itria: automations connected to the software the business already uses, and a Dashboard showing what has happened and what is waiting for a decision. Nothing goes to a customer without a person’s yes. On a test bench on 4 September 2026, with 30 documents including orders, delivery notes and invoices, the system extracted 234 fields out of 237 correctly, and the unreadable document deliberately placed in the batch was set aside without a single invented field.