# n8n vs Make pricing and hosting for European teams

Canonical URL: https://industrycontents.com/n8n-vs-make-european-teams/
Language: en-GB
Published: 2026-10-14
Structure updated: 2026-10-25

> Dated article snapshot. Canonical web page may receive later updates. Vendor figures retain their stated source-check dates. Illustrations are not product benchmarks.

Published: 2026-10-14 · Structure updated: 2026-10-25. Vendor prices and source checks retain the dates stated below.

## Quick answer

n8n bills complete workflow executions; Make generally bills module actions as credits. Compare the valid plan for the same trigger pattern, then add AI usage, hosting and maintenance before choosing.

## Key terms

**Workflow execution**: Workflow execution is a complete run of an n8n workflow, rather than an individual step.

**Make credit**: Make credit is Make’s billing unit; most ordinary module actions use one credit, while some AI features use more.

**Self-hosting**: Self-hosting is running the workflow platform on infrastructure your team maintains.

## In brief

n8n generally bills a complete workflow execution, while Make generally bills module actions as credits. Some Make modules are unbilled and some AI features consume more credits. Batching can reduce execution counts but does not establish a price saving by itself. n8n Community Edition avoids a cloud execution allowance while adding hosting and maintenance work; paid self-hosted plans have their terms. Managed hosting still leaves the buyer responsible for configuration, contracts and the data they process.

## n8n vs Make plans, cloud, free and self-hosted

n8n's [pricing page](https://n8n.io/pricing/) has an oddity. The self-hosted Community Edition has no execution cap. The hosted plans with a published price stop at 10,000 executions a month on Pro. The plan above that, Business at €667 a month on annual billing for 40,000 executions, is listed as self-hosted only, with a waitlist for a version n8n would host. The free route and the most expensive public route both end with a server that someone at your company runs.

Make has the opposite problem. It's sold as a managed service, so there's no server to run, yet [Make's GDPR page](https://www.make.com/privacy-and-gdpr) describes sub-processors, standard contractual clauses and a Celonis, Inc. registration under the Data Privacy Framework. A team that picks Make to avoid infrastructure work takes on vendor assessment instead.

## n8n executions vs Make credits, how billing works

n8n's pricing page defines an execution as one complete run of a workflow, whatever the number of steps or the volume of data in it. Starter is €20 a month on annual billing for 2,500 executions and Pro is €50 for 10,000, with unlimited users and workflows on every plan. For Business and Enterprise, n8n lists overage at €4,000 for each extra 300,000 executions and says runs continue when a quota is exceeded.

Make bills in credits. According to [Make's pricing FAQ](https://www.make.com/en/pricing), reading, searching, creating, updating, deleting and transforming data each cost one credit per action, as does iterating a row, while the Router and the error-handler modules (Rollback, Break, Resume, Commit and Ignore) cost nothing. Features that use Make's AI Provider can cost more than one credit. Our [Zapier vs Make vs n8n vs Gumloop pricing comparison](/compare-make-vs-zapier/) recorded Core at $12 a month for 10,000 credits and Pro at $21 on 14 October.

When credits run out on Make, scenarios stop. Incoming webhooks queue up to the plan's allowance and run once credits are added, and extra credits are sold in bundles of 1,000 or 10,000 at the price in the subscription.

| Comparison | n8n | Make |
| --- | --- | --- |
| Billed unit | One complete workflow run, called an execution | One credit per module action, more for some AI features |
| Iterating over rows | Inside the same execution | Iterating a row costs a credit, and each module after it runs once per row |
| Free of charge | Self-hosted Community Edition, under its licence | Router and error-handler modules, plus a Free plan of 1,000 credits a month |
| At the limit | Business and Enterprise runs continue and overage is invoiced | Scenarios stop until credits are added, and webhooks queue |
| Hosted data location | Frankfurt on n8n’s hosted plans, or wherever you host | Described on Make’s GDPR page, which links a data processing agreement and sub-processor list |

Caption: Two billing units, two cost curves. The table sets out Comparison, n8n, Make.

The table draws on the n8n pricing page and the Make pricing and GDPR pages, checked 14 October 2026. Prices are annual billing figures where the vendor page defaults to them.

## n8n vs Make: the same workload

Trigger frequency changes n8n’s execution count; Make’s credit count depends on the actions performed for each record.

Take 15,000 records a month, evenly spread across 30 days. Assume five billable, one-credit module actions per record on Make. This is a counting exercise, not a measured AI workflow: model usage, polling, pagination and retries sit outside the baseline.

| Design | Records per run | n8n parent runs | Make action baseline |
| --- | --- | --- | --- |
| One nightly batch | 500 | 30 | 75,000 credits |
| One hourly batch | About 21 on average | 720 | 75,000 credits |
| One trigger per record | 1 | 15,000 | 75,000 credits |

Caption: Illustrative arithmetic: 15,000 records × five one-credit actions. Parent-run counts assume one scheduled or event-triggered workflow, without additional production runs. These are unit counts, not equivalent bills.

Nightly batching cuts the n8n run count, but introduces up to a day's delay. Hourly batching offers a middle ground. Neither design is suitable just because the counter looks smaller: an urgent support escalation may need an immediate trigger.

The Make baseline stays flat only because every record still receives the same five actions. Filter records before expensive work and it falls; add enrichment calls, extra branches or a replay and it rises. Make documents that [replayed runs consume credits](https://help.make.com/scenario-run-replay). Read the actual scenario history before multiplying by monthly volume.

### Make retry costs

Make retry costs depend on which billable actions repeat, rather than a fixed percentage of the original bill.

Suppose 10% of records need all five actions repeated. That adds 7,500 one-credit actions to the Make baseline, taking it to 82,500 before the other exclusions. If only one action needs repeating, the addition is 1,500. Those failure rates are invented assumptions. The difference shows why recovery design belongs in the budget.

For n8n, count the additional production runs your recovery design creates rather than adding a percentage to every workflow automatically. Its [sub-workflow documentation](https://docs.n8n.io/build/flow-logic/break-workflows-into-smaller-parts) says sub-workflow executions do not count toward monthly execution limits. A separate top-level trigger and an internal sub-workflow are different cases.

### No magic threshold

There is no universal point where n8n becomes cheaper than Make. At 15,000 parent runs, n8n's listed 10,000-execution Pro allowance does not cover the example. At 75,000 Make credits, a 10,000-credit price cannot be used as the total bill. Get the valid tier or quote for each workload, use the same currency and billing period, and add model costs and maintenance. Our [cost-per-run guide](/guides/workflow-automation-pricing/) provides the wider method.

## AI steps add costs neither unit covers

n8n's Starter plan lists agents, MCP and AI nodes, plus AI models that run without an API key. Those features change costs once a workflow calls a model several times. The same page caps a single execution at five minutes on Starter and 40 minutes on Pro, so an agent loop could reach the time limit before it reaches the 2,500 execution cap. Multiple steps inside one execution do not themselves create a per-step platform charge, but new production triggers, runtime constraints and external services can change the total. Ask n8n where the models that run without an API key are hosted before sending personal data through them.

Make's [credit documentation](https://help.make.com/credits) distinguishes the connection type. With a custom provider connection, Make meters operations and the model provider bills tokens separately. With Make's AI Provider, credits reflect operations and tokens; automatic provider connections can include other usage factors too. Do not add a second model invoice to a scenario where the model cost is already included in its credits.

## n8n vs Make hosting and European data checks

n8n says data on its hosted plans is stored in Frankfurt, Germany, and that on self-hosted plans it sits wherever the customer hosts it, according to its [pricing FAQ](https://n8n.io/pricing/). Self-hosting therefore moves processing inside the company's perimeter and moves the obligations with it. Saved executions exist so users can inspect and debug runs, which means they can hold the data a run handled. The pricing page lists seven days of execution log retention on Starter, 30 on Pro and unlimited on Enterprise.

The outbound connections need a decision too. The [n8n deployment docs](https://docs.n8n.io/hosting/configuration/environment-variables/deployment/) show anonymous telemetry and version notifications switched on by default, and workflow templates loading from n8n's API. Turning telemetry off also disables the Ask AI feature in the Code node. Self-hosted Business and Enterprise licence keys must report to n8n's licence server every day, including the number of production executions, according to the pricing FAQ. In our reading a data protection officer would want those connections either switched off or listed in the record of processing.

Agencies need to check the licence. n8n's [Sustainable Use License](https://github.com/n8n-io/n8n/blob/master/LICENSE.md) allows use and modification only for internal business purposes or non-commercial use, and allows distribution only free of charge for non-commercial purposes. A consultancy running workflows for clients on Community Edition should read it before building a service on top.

Make's route is a contract. Its GDPR page describes a privacy programme run by Celonis, with an external data protection officer for Celonis SE, and links a data processing agreement and sub-processor list from its terms page. It says Celonis, Inc. is registered to the Data Privacy Framework and cites standard contractual clauses for sub-processors in countries without an EU adequacy decision. A hosted tool needs an Article 28 agreement and an entry in the record of processing, and the sub-processor list is where a transfer review starts. A third-party directory, [European Purpose](https://europeanpurpose.com/tool/make), says customers choose an EU data centre when they create a Make organisation. Make's GDPR page doesn't state that, so confirm the region in the account before loading personal data. Our [AI agent privacy and approval checklist](/guides/ai-agent-privacy-approval-checklist/) covers the approval questions that apply to either tool once agents are involved.

## Limits of the self-hosting case

A server invoice is only one line. Include backups, monitoring, upgrades, security patches and the time needed to restore a failed workflow. The earlier third-party hosting ranges are not a reliable production budget, so they have been removed from this calculation.

### n8n self-hosting labour costs

Use this monthly comparison: infrastructure + maintenance hours × internal hourly cost + model/API usage + paid licences. For an invented example, €15 of infrastructure plus two hours at €60 is €135 before models or licences. Against a €50 cloud subscription, that leaves an €85 gap before comparing capacity and features. These are budgeting assumptions, not hosting quotes or measured maintenance times.

Migration also has a bill. Estimate rebuild hours, authentication setup, test coverage and a period of running both systems. If a move costs €600 and saves €50 a month after maintenance, the simple payback is 12 months. At €10 savings it is five years. Cheap executions can come with a long receipt.

The execution advantage has a limit too. It exists when work is batched or when a workflow has many steps per run. A team that triggers one run per event pays for events, and Starter's 2,500 executions cover about 83 events a day.

n8n describes its telemetry as anonymous and its licence ping as usage data. Industry Contents didn't test what either sends.

## Test the awkward cases

An n8n or Make pilot should test duplicates, invalid inputs and API failures before the team relies on the workflow.

Before choosing, run a normal record, a duplicate, a malformed input and an external API timeout. Check whether recovery repeats a customer message or creates a second invoice. Log credits or executions, external charges, recovery time and whether the result passed review. Count successful business outcomes separately from successful-looking runs.

For a production workflow, name the person who handles a failed run and test restoring both configuration and credentials. Managed hosting reduces server work; it does not decide what to do with a duplicate payment or an incorrect AI classification.

## Matching each tool to the buyer

For a risk or legal lead, self-hosted n8n keeps processing on infrastructure the company controls and brings retention settings, patching, outbound connections and licence terms with it. Make keeps those with the vendor and replaces them with a processor agreement, a sub-processor list and a transfer review.

An operations lead should start with the trigger pattern. Batch-heavy or many-step workflows favour n8n's per-run unit. Short event-driven workflows favour Make's per-action unit, provided they stay short. Both buyers end up asking who in the company can run the server or sign the agreement.

n8n's paid ladder ends in a licence key that reports usage to n8n every day, and Make's ends in a vendor file with sub-processors attached. Each puts a different name in the record of processing. Your company needs to identify who answers for the workflow before comparing invoices.

## Method

The original pricing, licence and privacy review was conducted on 14 October 2026. On 25 October we rechecked n8n pricing, Make's pricing FAQ and credit rules, plus the linked sub-workflow and replay documentation. Make's current plan cards were not available in our retrieved page, so its earlier dollar prices remain dated snapshots. Workload, retry, labour and payback examples are our stated assumptions and arithmetic. We didn't run either tool, time executions or review the vendors' data processing agreements. This isn't legal advice.

## Frequently asked questions

### Is n8n cheaper than Make?

It depends on how workflows are triggered. n8n counts complete executions, so batching and long workflows may improve its economics, depending on the plan and other costs. Make counts a credit per action, so short event-driven workflows can cost less there. In the simplified example, nightly batching creates 30 n8n parent runs; per-record triggers create 15,000. Make starts at 75,000 one-credit actions before polling, retries and dynamic AI usage. These unit counts do not establish the cheaper valid plan.

### Is n8n GDPR compliant?

n8n says hosted data is stored in Frankfurt and self-hosted data stays wherever the customer hosts it. Compliance still depends on configuration and contracts, including retention settings, telemetry and, on hosted plans, the agreement signed with n8n.

### Can Make be self-hosted?

The European Purpose directory describes Make as a managed service only, and Industry Contents found no self-hosted edition on Make's pricing page. n8n's Community Edition is the self-hostable option.

### Is n8n free for business use?

The Community Edition is free to self-host for a company's internal business purposes under the Sustainable Use License. Hosted plans start at €20 a month on annual billing, and the self-hosted Business plan lists at €667 a month on annual billing.

### How should we compare costs before choosing?

Count your production triggers, steps, retries and model calls. Price the actual plan that covers that workload, then add hosting and maintenance time. Execution counts and action credits cannot be compared as if they were the same unit.

[Compare workflow billing across four platforms](/compare-make-vs-zapier/). Use the dated sources in this article for price context; the comparison method is an editorial framework, not a hands-on product benchmark.

For why orchestration may become more useful as models improve, see [the n8n example in our startup analysis](https://industrycontents.com/ai-startup-moat-foundation-models/). The commercial question is different from choosing the lowest-cost workflow plan. For a broader workload, the four-platform workflow comparison adds Zapier and Gumloop.

Depth update, 25 October 2026: [n8n prices and limits](https://n8n.io/pricing/), [Make billing FAQ](https://www.make.com/en/pricing), [Make credit rules](https://help.make.com/credits), [Make replay rules](https://help.make.com/scenario-run-replay) and [n8n sub-workflows](https://docs.n8n.io/build/flow-logic/break-workflows-into-smaller-parts). Earlier privacy and licence source dates below have not been refreshed.

## Sources

- [n8n pricing and FAQ](https://n8n.io/pricing/), [n8n licence file](https://github.com/n8n-io/n8n/blob/master/LICENSE.md) and [n8n deployment variables](https://docs.n8n.io/hosting/configuration/environment-variables/deployment/), checked 14 October 2026.

- [Make pricing and FAQ](https://www.make.com/en/pricing) and [Make privacy and GDPR page](https://www.make.com/privacy-and-gdpr), checked 14 October 2026.

- Third-party sources, not independently verified[European Purpose](https://europeanpurpose.com/tool/make), [Toolradar](https://toolradar.com/tools/n8n/pricing), [SyncGTM](https://syncgtm.com/blog/n8n-review) and [Automation Atlas](https://automationatlas.io/answers/n8n-pricing-self-hosted-vs-cloud-2026/).

IC Lab · AI Agent Pricing Tracker

Take the pricing rows into your budget.

Compare source-linked rows in the tracker. [Open the tracker and find the CSV export](/ic-lab/ai-agent-pricing-tracker/). The tracker offers a CSV export with a TalkHead signup. TalkHead is our weekly newsletter on AI in business.

**Related** [All comparisons](/ic-lab/compare/) · Zapier vs Make vs n8n vs Gumloop pricing · Workflow automation pricing · AI Agent Pricing Tracker

To inspect n8n alongside other self-hosted projects, browse [IC Lab’s GitHub repo directory](/ic-lab/github-repos/) and its setup and licence notes.

For the infrastructure behind increasingly capable AI agents, read our [analysis of Cloudflare MCP Code Mode and the future of agent tool calling](https://industrycontents.com/cloudflare-code-mode-mcp-explained/).