Scorecard
out of 10What matters most to you?
pick up to 3We'll score n8n and Make on just your priorities, using the side-by-side below.
Side by side
| Criterion | n8n | Make |
|---|---|---|
| How usage is counted | Monthly workflow executions, however many steps a run has (stronger) | Credits: 1 operation equals 1 credit for non-AI apps |
| Ease for a non-coder | Visual, but the community reports more technical setup | Visual drag-and-drop builder that the community finds easier (stronger) |
| Users included | Unlimited users on all plans | Unlimited users to share access across the organisation |
| Hosting and who runs it | Self-host on your own infrastructure, or use n8n Cloud | Hosted by Make on AWS in the EU or North America |
| Error handling | Error workflow that runs on failure, with email or Slack alerts | Error handlers keep scenarios running, with automatic retries |
| Licence for client work | Internal business use only, unless a separate agreement is signed | Not assessed in our research, so check Make's terms for client work before relying on this row |
The contenders in detail
n8n
- Best for
- Teams with someone technical to run it, and high volumes or data that must stay in-house
- Pricing
- Free Community edition, then paid plans billed per workflow execution
- Hosting
- Your own server or cloud account, or n8n Cloud
Pros
- Monthly executions are counted per run, however many steps a workflow has
- Community edition can be kept free indefinitely
- Can be self-hosted on your own server or in a private cloud
- Unlimited users, workflows and integrations on every plan
Cons
- Self-hosting means someone technical owns updates, backups and the server
- Features such as SSO and log streaming need a paid licence
- The free licence covers internal business use only, so client work needs a separate agreement
Make
- Best for
- Owner-run teams that want a visual, hosted builder and no server to maintain
- Pricing
- Plans billed in credits, with the cost of credits varying by plan
- Hosting
- Hosted by Make on AWS, in the EU or North America
Pros
- Visual drag-and-drop builder that the community finds easier for beginners
- Make hosts the infrastructure on AWS, so there's no server to run
- Error handlers keep a scenario running when one module fails
- Code app runs custom JavaScript or Python for advanced logic
Cons
- Credits add up as workflows gain steps and run more often, according to community reports
- On paid plans, custom AI provider steps are billed in credits, and you pay your AI provider for tokens
- Workflows run on Make's hosted AWS infrastructure, not on your own servers
You're looking at n8n vs make because your Make bill is climbing with every added step, but switching to n8n means running a server yourself. The choice comes down to who maintains the software and how you count usage: Make fits an owner-run team that wants full hosting handled for them, while n8n suits a setup where someone technical will keep it alive and your workflows are long enough that per-step billing stings.
How the two tools differ in practice
Both are visual workflow builders where you drag steps together on a canvas, so the interface won't feel foreign either way. The real split is who runs the software and how usage is counted. Make is a hosted service: it builds on AWS infrastructure located in the EU or North America, and that's the whole story from your side (at the time of writing).
n8n gives you more levers. You can use n8n Cloud, or self-host the Community edition on your own machine using Docker Compose if you're comfortable with that. If "self-host" is making you reach for a browser tab to check what it means, our plain-English explainer of n8n covers it without assuming you've deployed anything before.
How each one bills your workflows
This is where the two tools diverge, and why the bill can look very different for the same work.
The billing units
n8n counts monthly workflow executions, regardless of complexity. All plans include unlimited users, unlimited workflows, and every integration (at the time of writing). One run of a ten-step chain is one execution. Make bills in credits; it changed the term from "operations" to "credits", but for non-AI apps, 1 operation equals 1 credit (at the time of writing).
A quick illustration
Say you have a ten-step workflow: it checks an inbox, cleans the data, updates a CRM row, sends two follow-up emails and runs five other steps along the way. That's roughly ten operations. Run it 1,000 times in a month and:
- n8n counts 1,000 executions (10 steps don't add to the count)
- Make uses roughly 10,000 credits (one per operation, for non-AI apps)
That's an illustration using round numbers so you can see the ratio, not a price quote. Neither vendor gave us plan pricing in our research, so check n8n's pricing page and Make's pricing page for current figures before you commit to either.
Where AI steps sit differently
If a workflow calls an external model, the maths shifts. With Make, a custom AI provider connection (such as OpenAI or Claude) is billed in credits based on operations, and you pay your provider directly for tokens.
Who actually looks after a self-hosted n8n?
Here's the honest part that most "n8n is cheaper" threads skip. If you run n8n yourself, someone owns the server. That means someone owns the patches, uptime and backups. Make avoids this entirely: it hosts on AWS for you (at the time of writing), so there's no machine to keep alive.
What the free tier gives you
The n8n Community edition is free to use indefinitely, which is genuinely useful if you're just prototyping or running a handful of internal automations. Paid plans add features like SSO, environments, external secrets and log streaming through a licence key. The licence covers your own internal business purposes (and non-commercial use), so it's fine for a company running its own ops. It does not cover reselling n8n to clients; you'd need a separate agreement for that.
On the effort question, one forum poster wrote that Make is easier and more beginner-friendly visually, while n8n "can be self-hosted and has no limits, but it requires more tech setup".
What happens when a step fails
Error handling is where non-technical owners notice the difference most, because nobody reads docs at 9am when the invoice email isn't going out. The mechanisms are different, but both tools handle failure properly if you configure them.
n8n uses an error workflow: you point any workflow at a designated fallback that runs on failure. That workflow can send email or Slack alerts, and it's shareable across multiple workflows so you build the alert path once. Make takes a different route with in-scenario error handlers that keep the rest of the scenario running after a single module fails, plus automatic retries with exponential backoff for transient errors.
Neither approach is wrong; they just suit different mental models. If your team lives in Slack and wants one central "something broke" channel, n8n's model maps cleanly to that. If you'd rather each workflow absorb small hiccups without waking anyone up, Make's retry-and-continue behaviour fits better.
When each one wins (and where Zapier fits)
Make is the lower-risk default when nobody on your team wants a server in their life. You get unlimited users to collaborate with, a visual drag-and-drop canvas that people pick up quickly, and code steps if you need them: Make's Code App runs JavaScript or Python for the logic that pure blocks can't express (at the time of writing). It also hosts on AWS in the EU or North America, so there's no infrastructure to patch.
n8n wins when volume gets high enough that per-step billing is eating your margin. One Reddit user put it this way: "My workflows are growing, and the execution costs on Make are starting to add up." That sentence is the whole argument for n8n in a lot of cases. Unlimited users, unlimited workflows, one count per run. The catch is that someone has to be comfortable with Docker Compose or an equivalent deployment, because self-hosting isn't a toggle.
If Zapier is also on the shortlist, read the Zapier against Make comparison and the Zapier against n8n comparison. They cover the same billing logic from a different angle, and together with this page give you the full three-way picture before you pick.
What we'd tell a client to do
We build and host automations for clients, so our preference isn't pure vendor loyalty: it's whichever one your team can actually keep running. If nobody wants a server, start with Make; the hosted model means your only job is keeping the workflows accurate. If you have someone technical in-house, or if we're hosting it for you, n8n is the better fit for long workflows, because each run counts once however many steps it has.
Bring us your current automation bill and a list of what each tool runs today. We'll map which flows are worth moving, which ones stay put, and whether the effort is justified at your volume. If you want to see the shape of the work before booking anything, our workflow automation case study walks through a real build end-to-end, and the full scope lives under how we build and host automations.
In practice · Case study
Workflow Automation
Giving the editorial team their time back — permanently.
Read the case studyKey terms
- Execution
- n8n's billing unit: one complete run of a workflow, however many steps it has.
- Credit
- Make's billing unit. For non-AI apps, each operation uses one credit.
- Self-hosting
- Running the software on your own server or cloud account, not the vendor's.
- Error workflow
- A separate n8n workflow that runs when another workflow's execution fails.
Frequently asked questions
Is n8n free?
Is Make cheaper than n8n?
Can a non-coder use n8n?
Can I move from Make to n8n?
Can I run n8n for client work?
Sources
- n8n pricing, n8n (accessed 16 Nov 2026)
- Make pricing, Make (accessed 16 Nov 2026)
- Credits - Make Help Center, Make (accessed 16 Nov 2026)
- Compare editions - n8n Docs, n8n (accessed 16 Nov 2026)
- Host n8n - n8n Docs, n8n (accessed 16 Nov 2026)
- Community license - n8n Docs, n8n (accessed 16 Nov 2026)
- Handle errors gracefully - n8n Docs, n8n (accessed 16 Nov 2026)
- Error handling - Make Help Center, Make (accessed 16 Nov 2026)