Guide · Automation
How to use fewer credits in Make and fewer executions in n8n
Make and n8n count usage differently, so the way you build a workflow changes your bill. These tips come from each vendor's own documentation, with a few clearly marked exceptions.
Make: how credits are used
On Make, most apps use one credit per operation, and an operation is one module processing one bundle of data. Some exceptions: built-in AI modules also use credits based on tokens, and the Make Code app uses two credits per second of execution time. A few modules are free, including the error handler modules (Rollback, Break, Resume, Commit, Ignore), the scenario modules (Call a subscenario, Start a scenario, Return output, Customize run name) and Make Functions.
Make: tips that follow from that
- Watch what loops multiply. Make's own example: a Google Sheets search that returns 10 rows uses 1 credit, but a Delete Rows module after it runs 10 times and uses 10, so the scenario uses 11 credits in total.
- Aggregate before the expensive step. An aggregator uses one credit for each aggregation, however many bundles it combines. In another example from Make's documentation, adding a Text aggregator to combine contacts cut a run from 11 operations to 3.
- Be careful with iterators. An iterator uses one credit to split a list, but every module after it then runs once for each item.
- Choose the AI option on purpose. Make's built-in AI provider adds token-based credits to each operation. Modules that use your own provider are listed as a fixed one credit per operation, so compare both for your case.
- Filter early. This tip is common advice rather than something from Make's own pages: put a filter right after the trigger so later modules do not run for items you do not need.
- Check polling triggers. Third-party guides report that "Watch" triggers use a credit each time they check, even when nothing new is found, and that the default schedule is every 15 minutes (96 checks a day). I could not confirm this on Make's own pages, so check your scenario's run history. If it holds, an instant (webhook) trigger or a longer interval would cut the cost.
- Test with Run once while building instead of leaving a scenario on a schedule, and review your credit usage under your plan before you scale up.
n8n: what counts as an execution
On n8n's paid plans, only production executions count toward your quota. Those are runs started automatically by triggers, schedules or polling. An execution is one run of the whole workflow, so the number of steps or the amount of data in the run does not change the count.
According to n8n's documentation, these do not count: manual executions from the editor, sub-workflow executions (only the parent run counts), error workflow runs, and polling-trigger checks that return no data. Webhook triggers count one execution for every inbound request, even one with an empty body, but malformed requests that fail before the workflow starts do not count.
Some third-party articles say that manual runs and sub-workflows do count on Cloud. n8n's own documentation says they do not, so follow the documentation and check your usage page if numbers look off.
n8n: tips that follow from that
- Process batches in one run. Because a run counts once, handling many items in one scheduled run costs one execution. Sending one webhook per item costs one execution for each.
- Cut noisy webhooks at the source. Every valid inbound request counts, so avoid sending events you do not need to react to.
- Split big workflows into sub-workflows if it helps readability. Per the documentation, only the parent run counts.
- Test in the editor. Manual executions do not count, and n8n recommends keeping a workflow unpublished while you test it.
- Use polling triggers where they fit. A poll that finds nothing does not count.
- Consider self-hosting. The self-hosted Community edition has no execution cap, but you pay for the server and look after it. Our Automation page lists Nigerian providers.
Quick comparison
| Question | Make | n8n Cloud |
|---|---|---|
| What is counted? | Credits, mostly one per operation | Executions, one per production run |
| Do more steps cost more? | Usually yes, each operation uses credits | No, a run counts once |
| Does a list of 100 items cost more? | Often, if later modules run for each item | No, if handled in one run |
| Does testing cost? | Test runs use credits, so use Run once | Manual editor runs do not count |
This table is a simplification. Check the vendor pages below for exact rules.
Sources
- Make Help Center: How features use credits and Operations
- n8n documentation: Understand executions