Comparison
Make vs n8n: the billing units are not comparable
Most comparisons put the monthly prices side by side. That is the wrong comparison, because the two platforms are not selling the same unit.
No link on this page is an affiliate link. We are in the programme below, so we could earn a commission elsewhere on this site — but not from anything on this page. We recommend free tools where they are the better answer, including ones that earn us nothing.
- Make.com: 35% of subscription payments for 12 months; extra-operation purchases do not earn commission (terms verified 2026-09-24)
What we have and have not run
We have now run controlled tests on both platforms, plus Pabbly Connect, on 2026-09-22. Three scenarios each, against an endpoint we built and control, with every request confirmed in our own server logs. What follows is what those specific tests showed. It is not a general verdict on either product. We tested Make’s free plan and n8n’s Cloud free trial. We have never run n8n’s self-hosted edition, we did not test either platform at volume, and we did not test alerting on either.
The criteria, stated before the verdict
- Billing unit. What the platform actually charges for, which determines how your bill scales.
- Cost behaviour with multi-item work. Most real workflows process lists, and this is where costs diverge sharply.
- Free option. Whether you can solve the problem without paying at all.
- What running it costs you in time. Self-hosting is free in money and not free in effort.
Billing units
| Criterion | Make | n8n |
|---|---|---|
| Billing unit | credits renamed from “operations” | workflow executions on Cloud; unmetered when self-hosted |
| Does a 10-step workflow cost more than a 2-step one? | Yes — cost tracks the work done | On Cloud, one execution is one execution regardless of step count |
| Free option | Free tier with a monthly credit allowance | Community edition, self-hosted, no execution cap |
| What “free” costs you | Nothing, until you exceed the allowance | Your time: updates, backups, uptime and security are yours |
These units are not convertible. A credit, an execution and a task each measure a different thing, so the numbers in these columns must not be compared directly.
We ran the same workflow on a third platform as well. What one order cost on Make, n8n and Pabbly side by side sets out all three units, and what each of them charged for a run that failed.
DocumentedMake’s pricing page states a free tier of 1,000 credits per month, with Core at $10.59 monthly or $9 on annual billing. (Make.com, accessed 2026-09-22)
Documentedn8n Cloud bills by workflow execution rather than per step, and the self-hosted Community edition has no execution limit. (n8n, accessed 2026-09-22)
⚠ We could not verify thisWe could not verify current n8n Cloud pricing tiers from a primary source in a form we were willing to reproduce, so we have not put numbers in the table above for them.
Why this matters: measured line-item scaling
We built an order-to-spreadsheet workflow and measured how its cost behaves as an order gains line items. This is our own measurement of the workflow design, not of either platform:
| Line items | Units of work |
|---|---|
| 1 | 3 |
| 3 | 5 |
| 10 | 12 |
We tested it The relationship is units of work = line items + 2 (one fetch, one append per item, one summary notification). A ten-item order costs four times a one-item order.
This is the whole point of the comparison. On a platform that bills per unit of work, that multiplier lands on your bill and your customers control it. On a platform that bills per execution, it does not. Neither model is better in the abstract — but if your orders routinely carry many line items, the difference is large, and the pricing pages will not tell you.
Our verdict, and its limits
There is no overall winner here
These two platforms bill on different axes, and in our tests they failed in opposite ways. Which is cheaper or safer depends entirely on your workload, so the useful answer is a use-case one rather than a ranking.
The units do not convert. Make bills credits, n8n bills workflow executions. 13 and 1 are not comparable numbers — they measure different things.
Consider n8n when your orders carry many line items
In our tested 10-line workflow, n8n consumed 1 execution. A 1-line order also consumed 1 execution. The count did not move with order size. In the same scenario Make consumed 13 credits — one trigger, one fetch, one iterator and ten per-line writes — so its cost scaled with a number your customers control.
We measured Make at 10 lines but never at 1 line, so Make’s per-line slope rests on that single run plus its own operation counter, which reported “13 operations, 13 credits”.
Consider Make when you want a failure to stop early and cost less
We pointed both platforms at a destination that returned an error for every write. Make attempted one line item, stopped, and the failing run cost 4 creditsagainst 13 for the identical successful run. n8n attempted all ten and still consumed 1 execution — the same as a success. Neither wrote a single row, so neither left a half-finished mess at the destination.
What mattered more than either: nobody gets told
Neither platform alerted anyone by default. On both, the failure was visible only if someone opened the dashboard. Make showed no automatic re-execution and its Incomplete Executions queue was empty, because that storage is off by default. n8n showed no automatic re-execution in a verified 10 minute 56 second window, and its per-node retry is off by default. If you run either of these unattended, that is the thing to configure first.
In this specific test, n8n exposed the most debugging detail
Its failure view highlighted the failing node, visibly skipped the downstream nodes, showed the exact input item that failed beside the output, surfaced our destination’s own error text, and offered a “Debug in editor” replay. Make named the origin, included the response body and offered to build an error-handler route for us. Both were more informative than a small business would expect; we are describing one failure on one day, not ranking the products.
Two practical differences we hit while building
- Nested line items. Make accepted a nested path straight into its iterator. n8n split on the nested field directly. Neither fought us here — but Pabbly did, which is covered on its own page.
- Write ordering. In our tested 10-line workflow, n8n issued the writes concurrently and they arrived out of order. Make wrote them in sequence. If row order matters, n8n needs you to enforce it.
Who should avoid each
- Avoid Make if your orders routinely carry many line items and you need a predictable monthly bill, or if you need unattended recovery — in our test it did not retry.
- Avoid n8n Cloud if you want a permanently free hosted option: it is a trial, not a free tier. The genuinely free route is self-hosting, which we have not run and cannot vouch for.
What we have not tested
Native deduplication on either platform — our duplicate test measured how each behaved when our own destination rejected a duplicate with HTTP 409, not any built-in dedupe feature. Alerting configuration. Behaviour at volume or under concurrency beyond a ten-item fan-out. Retry behaviour with n8n’s per-node retry enabled or Make’s incomplete-execution storage switched on. Self-hosted n8n. Build times are not comparable either: we built Make by hand and imported n8n from JSON.
What we earn from this page
Nothing from this page. We are in Make’s affiliate programme and not in n8n’s — but no link on this page is an affiliate link, so this comparison earns us nothing either way. No vendor has seen this page or paid for anything on it, and we have earned no commission from any programme so far.
We are telling you this precisely because the asymmetry is uncomfortable: Make can pay us and n8n cannot. Make’s published rate is 35% for 12 months, n8n Cloud’s is 30% for 12 months, and we joined Make’s on 24 September 2026. Nothing above was reordered or reworded when we joined — the measurements and the conclusion are unchanged, and you can check that in our corrections log. Both rates are listed on our funding page. Self-hosted n8n Community — often the right answer for a technically comfortable reader — would pay us nothing even if we were approved.