n8n vs Make vs Zapier, which one should an SMB pick

n8n, Make and Zapier do the same job, connecting your applications and automating repetitive tasks. The choice therefore does not come down to the feature list, which looks very similar across the three, but to three far more practical questions. Zapier is the easiest to pick up and the most expensive as soon as volume climbs. Make offers the best power to price ratio for a team with no technical profile. n8n becomes the most cost effective when volume grows or when your data has to stay in house, in exchange for a little engineering.

Tandem is an n8n partner agency and deploys automations for small and mid-sized companies, from Doctolib to InfraVia through PayFit, Nomination and KparK. This article compares the three tools on what actually decides in a company, the billing rule, where the data is hosted, maintenance and the level of automation you are aiming for. Every price quoted was checked on the vendors' official pages on 8 August 2026.

n8n, Make and Zapier, what each one does

Zapier is the market pioneer and claims more than 9,900 connected applications as of 8 August 2026. Its logic stays deliberately linear, one trigger then a sequence of actions, which explains how easy it is to pick up. Make, formerly Integromat, offers a visual canvas where scenarios branch out with routers, iterators and fine grained error handling, for around 3,000 applications in its catalogue. n8n is an open source orchestrator, self-hostable, listing 2,014 integrations in its official catalogue and adding code nodes and native artificial intelligence nodes.

The number of integrations is impressive but rarely decides anything. All three tools can call any application programming interface through a generic HTTP node, which makes the real gap far smaller than the advertised one. What genuinely separates them is the way each vendor counts your consumption.

Execution, credit or task, the same automation counts three times

This is the least understood part of the three pricing pages, and the one that produces end of month surprises. n8n counts one execution per full workflow run, whatever the number of nodes crossed. Make counts one credit per module executed, trigger included. Zapier counts one task per successful action, leaving triggers, filters and paths free.

Take an ordinary automation, a submitted form that triggers eleven actions, enrichment, CRM creation, notification and archiving. Run 5,000 times a month, it consumes 5,000 executions with n8n, 55,000 tasks with Zapier and 60,000 credits with Make. The work performed is strictly identical. The meter varies by a factor of twelve.

This mechanism has a practical consequence. With Make and Zapier, adding a step to an existing scenario raises the bill in proportion to volume. With n8n, the same step is free. The exact rules are published by the vendors, both for Make credits and for Zapier tasks.

How much do n8n, Make and Zapier cost in 2026?

n8n is free in its self-hosted Community edition, with no execution limit. Its cloud plans start at 20 euros a month on annual billing for 2,500 executions, move to 50 euros for 10,000 and 667 euros for 40,000, prices checked on the official n8n pricing page on 8 August 2026. The full cost of a project, server and operations included, is covered in our article on the real pricing of n8n.

Make offers a free tier of 1,000 credits a month, then a Core plan at around 8 euros a month for 10,000 credits, a Pro plan at around 15 euros and a Teams plan at around 27 euros. Zapier gives 100 tasks a month on its free tier, but limits that tier to two step scenarios, with no filters and no paths. Its Professional plan starts at around 18 euros a month on annual billing for 750 tasks, and its Team plan at around 63 euros for 2,000 tasks. The Make and Zapier grids are published in dollars, so these amounts are orders of magnitude converted on 8 August 2026.

Read at entry level, this comparison flatters Make. Read at 50,000 units a month, it swings clearly towards n8n, because the execution meter climbs twelve times slower than the credit or task meter.

Where Louis Graffeuil ranks the three tools after two years of testing

Louis Graffeuil, founder of Tandem, publishes the Optimia newsletter every week. In the 7 November 2025 edition, he ranked the AI and automation tools he uses daily, from the most useful to the most disappointing. n8n sits in tier A, the must-haves, described as a flexible open source automation orchestrator. Make sits in tier B, the nice to haves. Zapier falls into tier D, the overrated, described as too limited and too closed.

That ranking deserves its nuance, and Louis gives it himself in the edition. It is a personal verdict, based on specific needs, not a benchmark. Zapier remains excellent for a small business that wants to plug two applications together in fifteen minutes without ever opening documentation. The too closed judgement is about something else, about the moment you want to step off the paved road, call an in-house application programming interface or run a piece of code in the middle of the scenario.

GDPR, hosting and data sovereignty

The shortcut that says n8n is the GDPR option and the others are not is simply wrong, and it costs time in board meetings. Make lets you choose your organisation's location between the European Union and the United States, and holds SOC 2 Type II certification. Zapier hosts in the United States and relies on standard contractual clauses. All three tools can therefore be used within a compliant setup, with the right contract in place.

The real question is a different one. It is about what actually travels. If your automations handle health records, credit files or investment documents, server location is not enough, because the content still crosses a third party's infrastructure. Only self-hosted n8n keeps the whole processing chain inside your own network, with your retention rules and your backups. That is the reason that comes up most often when our financial services clients settle on n8n.

n8n vs Make, which one should you pick?

n8n and Make resemble each other far more than they admit. Both offer a visual canvas, branches, loops and serious error handling. Make keeps the edge on the learning curve, on the quality of its ready made connectors and on entry cost. n8n keeps the edge on cost at scale, on hosting freedom, on code nodes and on orchestrating AI models.

• Pick Make if your team has zero technical profile, your scenarios stay under 10,000 runs a month and your data can live with a European vendor.
• Pick n8n if your volume passes a few tens of thousands of steps a month, if your automations call AI models in sequence, or if your data must not leave your infrastructure.
• Keeping both is a perfectly viable scenario. Several companies Tandem supports leave their marketing scenarios on Make and migrate only the heavy processing to n8n.

n8n vs Zapier, which one should you pick?

The trade-off between n8n and Zapier is sharper than between n8n and Make. Zapier targets the business user who wants a result in fifteen minutes and accepts paying for that simplicity. n8n targets the team that wants to keep control of the logic, the code and the hosting. Between the two, the price gap becomes spectacular beyond a few thousand actions a month, since Zapier bills every action while n8n bills every run.

One point deserves to be said plainly, because it often goes unsaid. Zapier remains unbeatable on the long tail of niche applications. If your automation depends on a little known business software, there is a fair chance a connector exists on Zapier and nowhere else. That single criterion sometimes settles the matter, and there is no shame in starting there.

Three questions that settle it in ten minutes

Here is the grid Tandem teams use in scoping workshops. It resolves most cases without any feature comparison, because it takes the constraints in the right order. The data constraint comes before interface comfort, and volume comes before the connector list.

One question comes even before that grid. Does your need really call for an automation, or would a simple assistant be enough? Louis details the four levels of optimisation available with AI in a dedicated article, and the field observation stays the same, many projects stop at level 2 with an excellent return on investment.

Does Claude Code replace n8n, Make and Zapier?

The question keeps coming up since development agents arrived. Louis Graffeuil put it bluntly on LinkedIn on 17 February 2026, with a deliberately provocative visual, then spent the whole post qualifying his own headline. A development agent like Claude Code reproduces the logic of a scenario very quickly, data aggregation, business rules, calls to application programming interfaces. What it does not replace is the infrastructure around the flow.

An automation platform brings four things that code alone does not. It deploys without multiplying micro-services, it monitors and alerts cleanly, it keeps connection credentials on the platform side and it makes building blocks reusable. Code, on the other hand, means deploying and maintaining, handling observability by hand and industrialising the infrastructure around the flow.

The trap is believing that a scenario running locally is a scenario running in production. When an automation breaks at 3am, nobody wants a terminal. What is needed is readable execution logs, automatic retries and an alert that reaches the right person. The two approaches therefore complement each other more than they replace each other, with Claude Code happily generating the logic and the custom nodes that the platform then executes.

The cost nobody budgets, maintenance

An automation is not a frozen deliverable. Application programming interfaces change, authentication tokens expire, data formats evolve and edge cases surface over the months. On the projects Tandem follows over time, one critical automation requires between 2 and 5 days of maintenance a year, whatever the tool. That figure weighs far more than the subscription gap between the three platforms.

That maintenance is not shared the same way. Make and Zapier absorb their connector updates and the operation of the service. Self-hosted n8n leaves you the database, the reverse proxy, the certificates, the backups and the version upgrades. An n8n project with nobody to hold those four lines ends up costing more than a Make subscription, in incidents rather than in euros.

n8n, Make or Zapier, our pick for a small business

For a French small or mid-sized company starting out, Make remains the best entry point, with a low cost, fast onboarding and European data centres. Zapier keeps its place when simplicity comes first and your applications are exotic. n8n becomes the right call as soon as one of three signals appears, a volume beyond a few tens of thousands of steps a month, data that must not leave your infrastructure, or AI processing at the heart of the workflows.

The worst scenario is still picking the tool before scoping the need. A two week AI audit identifies the automations worth building, sizes their real volume and makes the platform choice almost obvious. To discover n8n before deciding, start with our complete guide to n8n then with the step by step example of a newsletter automated with n8n.