A Zapier task, a Make credit and an n8n execution are different billing units. Comparing their advertised allowances as if they were interchangeable can produce a misleading estimate. Map the same workflow to each platform’s chargeable events before comparing a plan or a bill.
This guide uses current official definitions and a hypothetical workload. It does not claim that we deployed the workflow or found the cheapest platform.
What each unit counts
| Platform | Starting point for the count | Important qualification |
|---|---|---|
| Zapier | Successful chargeable action steps | Not every trigger or filter is an action task; consult the documented exceptions and recovery rules |
| Make | Credits consumed by module activity | Ordinary fixed usage defaults to one operation per credit, but some modules and AI paths use different accounting |
| n8n | Complete workflow executions | Count additional runs separately; execution allowance is distinct from other services and external costs |
The relevant definitions are in Zapier’s task-usage documentation, Make’s credits documentation and n8n’s pricing page. Make’s use of credits means that “operation” and “credit” should not be treated as universal synonyms.
A bounded example: 100 events, two actions
Assume 100 incoming events. For each event, the scenario creates a record and sends a notification. For this example only, both are successful chargeable actions in Zapier, and each is exactly one ordinary one-credit module operation in Make. In n8n, each event is processed in one complete workflow execution.
| Counted activity | Arithmetic | Result |
|---|---|---|
| Zapier’s two chargeable actions per event | 100 × 2 | 200 tasks |
| Make’s selected two ordinary module operations per event | 100 × 2 × 1 credit | 200 credits for that subset |
| n8n’s one complete run per event | 100 × 1 | 100 executions |
The Make row deliberately excludes ingress, polling and any other module operations. The n8n row excludes additional scheduled runs and retries. None of these rows is a complete invoice, and matching the first two numbers does not establish equal costs.
Adding a fifth chargeable action changes an action-based count. Adding another ordinary module operation changes the corresponding Make subset. Adding steps inside the same n8n run does not, by itself, mean five complete executions; the relevant unit remains the complete workflow execution described by the plan.
Build a quote from the workflow, then the plan
Record the expected number of events and how often each branch actually runs. A branch used by ten events should not be assumed to run for all 100. Then identify polling, scheduled activity, retries and the treatment of failure or recovery paths in the platform’s current documentation.
For AI activity, check the credit or token rules for the exact module and who bills the underlying model usage. An automation platform’s allowance does not necessarily include a separate provider’s charges.
Finally, apply the actual plan’s fixed fee, included allowance and applicable overage rules. Saved execution retention, logs and assistant features may also be separate plan considerations. Keep those alongside the workload count rather than hiding them in a headline “per automation” price.
For the broader operating tradeoffs, read the Make.com review and the Zapier-versus-Make decision guide. Their platform decisions still need a workload count before becoming a cost comparison.
Method and limits
The definitions were checked October 11, 2026. The worked example compares explicitly stated units, not complete production costs. Recheck the documentation when selecting a plan, and use an observed or carefully specified workflow record before attributing savings to a platform change.