Make, Zapier or n8n: what to choose for automation in your business

You have a form on your website, an invoicing program and a mailbox that receives a few dozen emails every day. Some of this work can be done without a person, but then comes the question of what to build it on. Three names come up most often: Make, Zapier and n8n.

In this guide I compare them on three typical automations from a small business. You will get a table showing how many billing units each automation uses at 1,000 and 10,000 runs, and a simple choice for three typical situations. I deliberately don’t give specific prices: price lists change often, so instead you get a formula to plug into the current price list.

Three automations I use to compare the tools

Tools are usually compared by feature lists. I am interested in what a specific job costs and how it works. That is why I take the three automations that come up most often in small businesses:

1. Form → CRM → SMS. A customer fills in a form, the data lands in a spreadsheet or CRM, and the customer receives a confirmation SMS. A notification for a salesperson would be a further step; I leave it out of the table.
2. Invoice → accounting. A cost invoice arrives by email as a PDF, the automation downloads the file, reads the data and sends it to an accounting program or to a folder for the accountant. Since 2026, invoices from most Polish companies are issued in KSeF (Krajowy System e-Faktur, Poland’s national e-invoicing system), so the data from them is usually easier to pull from KSeF through an accounting program. An automation with email and PDF is then useful mainly for foreign invoices and invoices from the smallest companies, which have to use KSeF from 1 January 2027.
3. Email → AI draft reply. An inquiry about an offer arrives in the mailbox, AI prepares a draft reply based on your price list, and you only check and send it.

Each of them is a few steps: a trigger (something happens) and actions (something gets done). How the tool counts these steps determines the bill more than the price of the plan itself.

How each tool counts operations

This is the most important difference, and the easiest to miss in price lists. I describe it according to the providers’ price lists and help pages (as of 20 October 2026). Before buying, check it in the current price list, because providers change names and limits.

• Zapier counts tasks. A task is every successful action step. The trigger, filters, Paths, Formatter and steps that end in an error don’t count, and some AI steps may use more than one task. So a simple automation with two actions uses two tasks per run.
• Make counts credits (until 27 August 2025 this unit was called an operation). Every module in a scenario, including the trigger, usually uses one credit per run. The router and filters use no credits, and AI modules using Make AI Provider may use more, depending on the number of tokens. A scenario with three modules is three credits per run.
• n8n in the cloud version counts executions: one run of a workflow is one execution, regardless of the number of steps. The free Community version on your own server has no execution limit; you pay for the server. The paid Business plan, however, has an execution limit, even though it also runs on your own server.

Conclusion: the more steps an automation has, the more n8n pays off. The simpler the automation and the less often it runs, the smaller this difference.

Table: how many units you will use at 1,000 and 10,000 runs

I assume the following numbers of steps. This is an example; your workflow may have more or fewer:

• automation 1: trigger (form), save to CRM, send SMS (3 steps),
• automation 2: trigger (email with an invoice), download the attachment, read the data, save in accounting (4 steps),
• automation 3: trigger (email), request to AI, save the draft in the mailbox (3 steps).

Below is the number of billing units according to the rules from the previous section:

How do you work out the cost from this? Take the current price list of the chosen tool, find a plan that fits your number of units, and add the price. For orientation, according to price lists as of 20 October 2026: Zapier’s free plan gives 100 tasks a month, Make’s free plan 1,000 credits, and the cheapest paid n8n Cloud plan (Starter) 2,500 executions. For example, with a pool of 1,000 units a month, the “form → CRM → SMS” automation fits in it about 500 times in Zapier, about 333 times in Make and 1,000 times in n8n.

The most common mistake is comparing plans by the label “1,000 operations” in each tool. It is not the same unit, so it is not the same amount of work.

The table assumes that each run concerns one inquiry or one email and that the trigger reacts immediately (e.g. through a webhook). In Make, a trigger that checks a mailbox on a schedule uses a credit with every check, even when it finds no new email. Checking every 15 minutes is about 2,900 credits a month before the automation does anything. Zapier does not count tasks for such checking. The AI step in automation 3 may, in turn, cost more than one unit in Zapier and Make, depending on how you connect the model.

Customer data and GDPR: where it runs

Each of these automations passes personal data through: a name, a phone number, the content of an email, invoice data. For you as the data controller this means you must know where the data is processed and whether you have a data processing agreement with the tool’s provider. I describe the state as of 21 October 2026 according to the providers’ documents. This is not legal advice, so if in doubt, consult a lawyer or a data protection officer about the data processing agreement.

• Make lets you choose the data center region: the EU or the US. You choose it when creating the organization and, according to Make’s help, it cannot be changed later, so for customer data choose the EU right away. Even in the EU region, some AI features (e.g. the Maia assistant and OpenAI models in Make AI Provider) are processed in the US. Make provides a data processing agreement (DPA) together with its terms.
• Zapier, according to its privacy page, stores data, including your customers’ personal data, on AWS servers in the US and does not offer EU-only data storage. It bases transfers of data to the US on EU-US Data Privacy Framework certification, and its data processing agreement (DPA) contains standard contractual clauses. Zapier does not sign data processing agreements prepared by customers.
• n8n in the cloud, according to n8n’s price list, stores data in the EU, on servers in Frankfurt. You can also set it up on your own server in the EU, for example on a VPS in Poland, Germany or Finland. Then the execution history and the credentials for services stay on your server, although the automation itself still passes data on, e.g. to a CRM, an SMS gateway or an AI provider. This is a good route when an automation handles a lot of personal data or a customer requires EU hosting.

There is one more element: if an automation includes AI, the content of the email goes to the model provider. I write about choosing a provider and the data processing agreement in the guide ChatGPT or Gemini for business. I also keep customer data with me in the EU as part of the Customer data in the EU service, when a project requires it.

Self-hosted n8n: license and hidden costs

Self-hosted n8n sounds like free automation with no limits. In practice that is half true, so check two things.

License. The self-hosted version of n8n (Community Edition) is free, but it runs under n8n’s own license (the Sustainable Use License), not a classic open-source license. According to the license FAQ, you may use it in your own business for your own processes, and also build and maintain workflows for clients for a fee, as long as the clients do not create or edit them on your instance themselves. Without a separate agreement with n8n, however, you may not offer n8n as a service in which clients build their own workflows, or use Enterprise features that require a license key. Before deploying, read the current text of the license, especially if you also build automations for your own clients.

Costs beyond the plan price. A server costs a few dozen zlotys a month, but someone has to:

• update n8n and the system,
• make backups of workflows and data,
• secure access (password, HTTPS, address restrictions),
• react when an automation stops working.

If you don’t have such a person in the company, include that time or external support in your calculation.

Which to choose: 3 situations

Situation 1: a few simple automations and no IT person

You have a form, a spreadsheet and an email confirmation. There are a few hundred runs a month, and the data is mainly a name and a phone number. Choose Zapier or Make on a ready-made cloud plan. You can set up the automation in one evening and don’t have to install anything. Zapier is usually simpler to start with; Make offers more possibilities for complex scenarios.

Situation 2: many steps, many runs

An invoice automation has four or five steps and receives thousands of documents. In Zapier and Make every step is a billing unit, so the bill grows fast. Here n8n pays off, in the cloud or on your own server, because one run is one execution regardless of the number of steps.

Situation 3: customer data, GDPR requirements and AI

An automation reads customers’ emails, sends the content to an AI model and saves a draft. Personal data passes through several systems. Here the best choice is n8n on a server in the EU, with control over what you send and where. For such deployments, it is worth adopting a rule: AI prepares the draft, and a person sends it.

What not to automate and what mistakes businesses make

Automation doesn’t always make sense. Before you build anything, check this:

• A process you can’t describe. If people in the company do it four different ways, agree on one first.
• A task done three times a month. Building and monitoring the automation will take more time than the manual work.
• Sending without review. An automation can send a wrong amount, a wrong deadline or a reply to spam. At the start, keep human approval.
• No error notification. An automation that quietly stopped working is worse than no automation. Set up an alert to an email address.
• No documentation. Write down what the automation does, where the passwords are and who looks after it.

If your automation is to send SMS messages (confirmations, reminders, a code), I connect the gateway in the SMS API service, together with message templates. For automations with AI in company processes, from quotes to handling conversations, there is AI implementation in your business.

Summary

For a few simple automations choose Zapier or Make in the cloud, and for many steps, a large number of runs and customer data consider n8n, preferably on a server in the EU. Before you buy anything, count the billing units using the table and check the current price list and the license.

If you prefer that I build and test such an automation, see process automation: I will do one process from PLN 1,490 net. Write to me, describe what you do by hand today, and within 24 hours you will get a quote and a tool suggestion.