n8n is for teams with someone who can read JSON without flinching, and who either want their data to stay on their own servers or want to drop code into a workflow when the visual nodes run out of road. Make is for teams who want a cloud service that never asks them to think about a server, and are willing to learn a builder to get real branching logic.
Our Zapier vs Make comparison covers the easy-versus-capable end of this market. This is the other axis: who runs the thing, and how the bill is counted. Those two questions decide more automation purchases than feature lists do.
The short answer
- Pick n8n if data residency or a contract clause means the automation cannot run on someone else’s infrastructure.
- Pick n8n if your workflows are long — twenty, thirty, fifty steps — because it bills per execution, not per step.
- Pick Make if nobody on the team wants to own a server, patch it, back it up and monitor it.
- Pick Make if your workflows are short and frequent, and you want a $12-a-month entry point.
When neither is the right answer
Check what you already pay for first. On Microsoft 365, Power Automate may already sit inside licences you hold. On Google Workspace, Apps Script covers a surprising amount of scheduled data-shuffling for nothing. Both are less pleasant than a purpose-built canvas, and both remove a line item.
Check the two apps themselves next. A great many automations exist only because nobody looked for the native integration. Most CRMs talk directly to most email tools; most accounting packages talk directly to most payment processors. A direct connection has one vendor, no credit meter and no middle layer to debug at 2am. And if one job needs to run on a schedule and someone technical is on staff, a script and a cron entry is free and takes an afternoon. The reason to buy an automation platform is the tenth workflow, not the first.
If the honest situation is that nobody will maintain any of this, do not start. An unowned automation platform becomes a set of silent failures that corrupt data for months before anyone notices. That is worse than a manual process someone does badly on Fridays.
What n8n is genuinely better at
Running on your own infrastructure. The Community Edition is free, published on GitHub and runs in a container on hardware you control. For healthcare, finance, legal and European organisations with data residency obligations, this is not a preference, it is the entire reason the product is on the shortlist. Make has no equivalent.
Code where visual building gets awkward. Custom JavaScript or Python steps sit inside the same workflow as the visual nodes, so the usual no-code wall — the transformation that is one step too fiddly for the interface — simply is not there. The same openness makes it a common choice for assembling AI agents, with nodes for models, vector stores and human-in-the-loop approvals.
Billing that does not punish long workflows. n8n counts workflow executions with unlimited steps. A fifty-node workflow costs the same as a three-node one. If your automations are elaborate, this changes the total by an order of magnitude.
The trade-off: free self-hosting is not free. You pay in servers, upgrades, backups, monitoring and security patching, and for a small team the cloud plan is usually cheaper once you cost your own hours honestly. It is fair-code rather than open source — the Sustainable Use License permits internal use and modification but restricts reselling n8n itself as a hosted service. Single sign-on, environments and advanced logging sit in paid tiers. And the learning curve is real: expressions, the data structure passed between nodes and error handling take genuine time.
What Make is genuinely better at
Being somebody else’s problem. No server, no upgrade window, no patching, no backup policy, and a working paid plan for twelve dollars a month. For a business with no technical staff, that is worth more than any feature.
Visual debugging. After a run you open each module and inspect exactly what went in and what came out. Tracking down a broken scenario is faster than reading logs, and this is consistently the thing Make users say converted them.
Real logic without code. Routers split a scenario down multiple paths, filters stop records that do not qualify, and iterators and aggregators loop over lists and reassemble results — all in the interface, no JavaScript required.
The trade-off: billing counts module operations, so a badly filtered trigger or a loop over a long list burns through the allowance quickly — filter early in the scenario, not late. The free plan is a trial in practice: two active scenarios, a fifteen-minute minimum interval between runs, a five-minute execution cap and a 5 MB file size limit. Arrays, bundles, iterators and aggregators trip up almost every new user. And connector depth varies — some app modules cover an API thoroughly, others expose only common actions, so confirm the operation you need exists before you design around it.
Pricing, checked August 2026
Make quotes in US dollars, with “Pay monthly” selected by default. Free is $0 per month with up to 1,000 credits per month. Core is $12 per month for 10,000 credits per month. Pro is $21 per month for 10,000 credits per month. Teams is $38 per month for 10,000 credits per month. Enterprise is custom-priced. Annual billing is advertised as saving 15% or more. All plans include unlimited users. Credits scale upward on each tier, so those figures are the entry point for each plan rather than the only price.
n8n quoted us in euros when we checked. Starter is 20€ per month billed annually and includes 2,500 workflow executions with unlimited steps. Pro is 50€ per month billed annually with 10,000 executions. Business is 667€ per month billed annually with 40,000 executions and a self-hosted option. Enterprise is priced on enquiry. The page carries a “Monthly” and “Annually (Save 17%)” toggle; the monthly rates were not displayed on the cards we read. All plans include unlimited users and unlimited workflows. The self-hosted Community Edition is free.
The unit is the whole comparison, and it is easy to get wrong. Make charges roughly one credit per module action, so a scenario with twenty modules that runs a hundred times a month consumes around two thousand credits. n8n charges one execution regardless of node count, so the same workflow costs a hundred executions. Short scenarios that fire constantly favour Make. Long workflows that run occasionally favour n8n by a wide margin. Model your actual volume against both units, because the headline numbers are not comparable.
This comparison is compiled from both vendors’ published documentation and pricing pages. We have not tested either product hands-on.
Migration and lock-in
Automation platforms lock you in through accumulated undocumented logic, not through data.
n8n has the better portability story, and it is structural rather than generous. Workflows are JSON files you can export and import, and the same workflow runs on n8n Cloud or on your own container. You can move between hosting models without rebuilding, and keep workflow definitions in version control alongside your other code. Make scenarios also export as JSON blueprints, but only into Make.
Neither exports credentials, and neither should. Whichever you leave, every OAuth connection, API key and webhook endpoint has to be re-established by hand at the far end. Inventory those webhooks before you switch, not after.
The n8n-specific risk is the self-hosting bargain. Free until the person who set it up leaves, at which point you own an unpatched container running your billing integrations and nobody knows the upgrade path. If you self-host, write down where it runs, who patches it and how it is backed up, on day one.
The Make-specific risk is the meter. Bills grow quietly as scenario volume grows, and the fix is almost always redesign rather than a bigger plan — moving a filter earlier, aggregating before an API call. Review your highest-consuming scenarios quarterly.
Side by side
| Decision criterion | n8n | Make |
|---|---|---|
| Self-hosting | Yes — free Community Edition | No |
| Billing unit | Workflow executions, unlimited steps | Credits, roughly one per module action |
| Free tier (Aug 2026) | Self-hosted Community Edition | $0/month, up to 1,000 credits/month |
| Entry paid tier | Starter, 20€/month billed annually, 2,500 executions | Core, $12/month, 10,000 credits |
| Mid tier | Pro, 50€/month billed annually, 10,000 executions | Pro, $21/month, 10,000 credits |
| Upper published tier | Business, 667€/month billed annually, 40,000 executions | Teams, $38/month, 10,000 credits |
| Currency shown | Euros when we checked | US dollars |
| Licence | Fair-code (Sustainable Use License) | Proprietary SaaS |
| Main failure mode | An unpatched server nobody owns | A credit bill nobody is watching |
The recommendation
Regulated business, or one with a data residency clause: n8n, self-hosted. It is the only one of the two that can satisfy the requirement, so the rest of the comparison is academic. Budget for whoever will patch it.
Small business with no technical staff: Make. The $12 Core plan is a fair price for a working automation platform, and you will never think about a server. Self-hosting n8n would cost you more in salvaged weekends than the subscription saves.
Agency building automations for clients: n8n, on cloud rather than self-hosted. Long multi-step workflows are your normal output, and per-execution billing is dramatically cheaper than per-action at that shape. Read the Sustainable Use License first if you plan to resell hosting.
Ecommerce team syncing orders and stock: Make. Short, frequent, well-supported connectors, and the visual debugger earns its keep when an order fails to sync during a sale.
Anyone with three simple two-app automations: neither yet. Use the native integrations, or Zapier if you want it done by lunchtime.
FAQs
Is n8n open source?
Not in the standard sense. It is fair-code, under the Sustainable Use License, which permits internal use and modification but restricts reselling n8n itself as a hosted service. The source is public on GitHub and you can self-host for free. If you intend to build a commercial product on it, read the licence text rather than the summary.
Which one is actually cheaper?
It depends on the shape of your workflows, not their number. Count the average steps per run and the runs per month. Multiply steps by runs for Make’s credit consumption; use runs alone for n8n’s executions. Long workflows favour n8n, short frequent ones favour Make. Do this before you look at the pricing pages.
Plan limits, credit and execution allowances, currency display and feature gating change regularly in both products. Make’s figures above were taken from its own pricing page in August 2026 in US dollars; n8n’s were taken from its own pricing page in euros on the same date. Check both in your own currency before committing.
