Framing

All three platforms solve the same basic task: connecting systems without anyone writing custom code for every connection. Triggers (e.g. a new email, a new record) start a chain of actions (e.g. create an order, send a message, update a spreadsheet). The difference lies in operating model, data location and billing.

You use Zapier and Make as the vendor’s cloud service. Make was formerly called Integromat. n8n is available as a cloud service and for self-hosting on your own or rented servers.

Comparison table

Vendors’ list prices and statements, as of 21 October 2026. Sources at the end of the page.

Criterionn8nZapierMake
Billing unitexecution: one run of the whole workflow, regardless of the number of steps; for AI features the pricing page lists separate credits (“Assistant credits”, “Gateway credits”)task: every successfully completed action; polling for new data and built-in tools such as Filter do not count; AI steps, code and some apps can use more taskscredit: most module actions 1 credit, some AI features more; each check of a trigger for new data also uses 1 credit, according to Make’s help pages; router and error handling do not count
Entry per pricing pagefree tier: pricing page shows a trial (1,000 executions); smallest paid cloud plan: Starter, €20/month billed annually, 2,500 executions/monthfree tier: Free, 100 tasks/month; smallest paid cloud plan: Professional, $19.99/month billed annually, 750 tasks/monthfree tier: Free, 1,000 credits/month; smallest paid cloud plan: Core, $9/month billed annually, 10,000 credits/month
Cloud data locationstorage location: EU, according to the pricing page servers in Frankfurt; built-in AI features: partly run via providers outside the EU; transfer basis per the vendor: Standard Contractual Clauses in the data processing agreementstorage location: AWS servers in the US, Zapier does not offer EU-only storage; built-in AI features: partly run via providers outside the EU; transfer basis per the vendor: EU-US Data Privacy Framework and Standard Contractual Clauses in the data processing agreementstorage location: the organisation runs in a data zone (US1, US2, EU1, EU2); built-in AI features: partly run via providers outside the EU; transfer basis per the vendor: depending on the contracting entity, the Data Privacy Framework (Standard Contractual Clauses as a fallback) or Standard Contractual Clauses (per the data processing agreement)
Self-hostingpossible: community version on GitHub (Sustainable Use License); Business plan self-hosted only, €667/month billed annually (40,000 executions)vendor’s cloud servicevendor’s cloud service
Operation and updatescloud: with the vendor; self-hosted: servers, updates, security and backups with you or your IT providerwith the vendorwith the vendor

Which platform fits when

n8n fits if your workflows have many steps, since billing is per execution. And if storage in the EU or self-hosting is an option for you.

Zapier fits if your workflows often check for new data, since polling does not count as a task. And if storage in the US is an option for you.

Make fits if your workflows have many branches (routers), since routers do not count as credits. And if a data zone in the US or in the EU is an option for you.

AI features and self-hosting: where data goes

This applies to all three platforms: if a workflow or a built-in AI feature calls a language model or another cloud service, that data goes to the respective provider, regardless of where the platform itself runs. Check the platform’s sub-processor list for this and, for your own API connections, the model provider’s privacy terms or data processing agreement.

Self-hosting means you or your IT provider run the server, apply updates and secure it. It does not automatically mean no data leaves your premises. More on this in the article GDPR-compliant AI providers for SMEs.

Sources

All retrieved on 21 October 2026: