Make vs n8n

If a marketing, sales or operations team will build and change the automations itself, choose Make. Its visual editor is the easiest to read, it has more ready-made connectors, and there is nothing to host.

If you have a developer nearby, handle sensitive data, or run tens of thousands of executions a month, choose n8n: you can host it yourself, write JavaScript or Python inside a workflow, and you pay per execution rather than per step.

Between the two, the deciding question is rarely a feature. It is who will maintain the automations once we have handed them over.

Make, formerly Integromat, is a visual automation platform. You build a scenario by placing modules on a canvas: a trigger, then the steps that read, transform and send data between your tools. Routers split a flow into branches, filters decide what continues, and iterators loop over lists, all without writing code.

Make runs only in its own cloud, with European and American regions. You pay for a monthly volume of operations, and every module that runs counts. That model is cheap for short scenarios and gets expensive for long ones that run often. Its strengths are the size of its catalogue and how readable a scenario stays for someone who did not build it.

n8n is a workflow automation tool released under a fair-code licence. On screen it looks like Make, with nodes on a canvas, but it is built for teams that want control. You can use n8n Cloud, or install the community edition on your own server and keep every piece of data inside your infrastructure.

n8n bills per workflow execution, not per step: a workflow with forty nodes costs the same as one with three. Code nodes accept JavaScript and Python, any API is reachable through the HTTP node, and its AI nodes let you build agents with the model of your choice, including one you host. The trade-off: it asks for more technical comfort, and if you self-host, updates and backups are yours.

Nine criteria we look at on every automation project. The last column says which tool has the edge, from our experience on client work.

The two tools do not count the same thing, and that matters more than the price grid. Make counts operations: each module that runs is one. A scenario with eight modules that runs 50,000 times a month uses about 400,000 operations. n8n counts executions: the same workflow uses 50,000, whatever its number of nodes. Self-hosted, the community edition has no execution limit; you pay for the server instead.

In practice, Make is usually cheaper for a handful of short scenarios, and n8n becomes cheaper as workflows get longer and volumes grow. Our work, setup and maintenance, costs the same on both, so the platform line is what tips the balance. We read both pricing pages before every estimate and date the figures we give you.

Make processes your data in its cloud. You can choose a European region and sign its data processing agreement, and for most marketing and sales flows that is enough. When the data is sensitive (health, finance, HR, or customer records your security team wants to keep in-house), self-hosted n8n is the simpler answer: workflows, credentials and execution logs stay on servers you control, in the region you choose.

Self-hosting is not free of work. Someone has to apply updates, watch the server, rotate credentials and back up the database. When we host n8n for a client, that is part of the maintenance plan; when the client hosts it, we document every step so their team can run it.

Both platforms now let you put a language model inside an automation: classify an email, summarise a call, draft a reply, extract fields from a PDF. Make offers AI modules and agents that are quick to set up and work well for single, well-defined tasks.

n8n goes further when the agent has to reason over several steps. Its agent nodes handle memory, tools and retrieval from your own documents, and you choose the model, from Claude or GPT to an open model running on your server. For a support assistant grounded in your knowledge base, or an agent that reads a document, checks your CRM and drafts an answer, we usually build on n8n or in code.

Make has the larger catalogue of ready-made connectors, and each one exposes many actions without configuration. For common business tools (CRM, email, spreadsheets, forms, invoicing) you will rarely need anything else.

n8n has fewer native integrations, but its HTTP and code nodes reach any API, and you can write the exact transformation you need in JavaScript or Python. When a stack includes an in-house tool or a niche ERP, that flexibility saves more time than a longer connector list.

This is the question we ask first. If the people who will change the automations next month work in marketing or operations, Make’s editor is the one they will read without help, and the platform takes care of uptime. If a developer is around, or if we maintain the automations for you, n8n gives more room: version control, tests, reviews, and separate environments for staging and production.

Either way, an automation is a small piece of software. It needs an owner, a log you can read, and alerts when it fails. We set those up on both platforms.

What we chose on recent client projects, and why.

Teams usually move when their Make bill grows faster than their volume, or when security asks for self-hosting. Scenarios are rebuilt, not imported, so we plan a migration like a small project.

Our notes from the projects we run on each platform. The full details are in our Make and n8n reviews.