A8gent
GlossaryAutomation mechanics

What is execution vs task vs operation?

Also called: automation billing units, credits vs tasks vs operations

Executions, tasks, operations and credits are the different units automation platforms bill in - and because one workflow run can be one execution, three tasks or twelve operations, the same automation can cost wildly different amounts on different platforms.

Written by A8gent
Contents
  1. At a glance
  2. Why this is the most expensive thing people get wrong
  3. How each unit is actually counted
  4. How to compare properly
  5. What you should know
  6. The unit-by-unit pages
  7. FAQ
  8. Related terms

At a glance

Category
Automation mechanics
n8n - execution
One full workflow run, regardless of node count, data volume or duration
Zapier - task
Any successful action; triggers, filters, paths and failed steps do not count
Make - operation
One module run per bundle processed; trigger modules run once regardless of bundles
Credits
Vendor-defined abstraction, often varying by feature or model tier - check the rate card
Biggest cost swing
Loops - hundreds of billed units on per-action platforms, free on per-execution ones

Why this is the most expensive thing people get wrong

Every automation platform bills in its own unit, and the units are not comparable. One workflow that runs once and touches eight steps might be billed as one execution on one platform, three tasks on another and twelve operations on a third. Comparing the monthly sticker price without translating the unit is the single most common costing mistake in automation projects, and it can be wrong by an order of magnitude.

The units also change what you should build. When you pay per step, you consolidate steps and avoid loops. When you pay per run, step count is free and you care about how often the workflow fires. The same requirement produces genuinely different designs on different platforms.

How each unit is actually counted

Execution (n8n). n8n bills per workflow execution: one run of a whole workflow is one execution regardless of how many nodes it contains, how much data it processes or how long it takes. A two-step workflow and a two-hundred-step AI agent both count as one. That makes complex workflows cheap and frequent triggers expensive.

Task (Zapier). Zapier's documentation defines a task as any successful action that runs. Trigger steps do not count. Filter and Paths steps do not count. Failed or halted actions do not count, and neither do several utility steps such as Formatter, Delay, Looping and Storage. So a Zap with a Gmail trigger, a Dropbox upload and a Slack message uses two tasks per run. Some products are priced differently again - as of August 2026 Zapier documents Code by Zapier at one task per run and Zapier MCP at two tasks per successful tool call.

Operation (Make). Make defines an operation as a single module run to process or check data. Each module in a scenario consumes operations according to how many bundles it processes, with the exception of trigger modules, which run once regardless of how many bundles they return. The consequence is multiplication: if a trigger returns ten bundles, each following module performs ten operations. A five-module scenario processing ten records is fifty operations, not five.

Credit. Used by several AI-first platforms as an abstraction over compute and model usage. Credits are the least comparable unit because the conversion rate is vendor-defined and often varies by feature or model tier - check the current rate card rather than assuming, since these change frequently.

How to compare properly

Model one real workflow. Write down the trigger frequency per month, the number of steps, and how many records a typical run processes. Then apply each vendor's own counting rules and multiply out. On a high-frequency, few-step workflow n8n's per-run model can be expensive relative to per-action pricing. On a low-frequency workflow that loops over hundreds of records, per-operation pricing multiplies alarmingly and per-run pricing wins.

Watch the loops especially. A step inside a loop over two hundred rows is two hundred billed units on per-operation and per-action platforms and free on per-execution ones. That single fact decides more platform choices than feature lists do. All the figures above are as published in August 2026 - vendors revise both prices and counting rules, so verify against the current documentation before you commit.

What you should know

Design to the unit you are paying for. On per-action platforms, combine steps, filter as early as possible so cheap filter steps discard work before expensive action steps run, and avoid processing records one at a time when a batch call exists. On per-execution platforms, do the opposite: batch triggers so one run handles many records rather than firing a hundred times.

Also budget separately for model tokens. The platform unit and the AI provider bill are different lines, and an agent step that costs one execution can still cost real money in tokens. Our workflow cost calculator and AI automation cost page cover both sides.

The unit-by-unit pages

This entry is the translation layer. Each unit has its own page with the vendor's current counting rules, allowances and expiry policy, all checked in August 2026: Zapier task, n8n execution, Make operation, Latenode CPU second, Gumloop credit, Voiceflow credit, Bardeen credit, Copilot Credit, Fin resolution, Power Automate licensing units, Workato recipe and Stack AI run.

Reading them side by side makes the point better than any comparison table: there are at least five genuinely different billing philosophies in this market - per action, per run, per row, per second of compute, and per licensed seat or process - and two of the vendors above do not publish a price at all.

Execution vs task vs operation FAQ

What is the difference between an execution, a task and an operation?

They are three vendors' billing units for the same thing. n8n counts one execution per full workflow run whatever it contains. Zapier counts one task per successful action step. Make counts one operation per module run per bundle. The same workflow can therefore cost very different amounts.

Does a Zapier trigger use a task?

No. Zapier documents that trigger steps do not consume tasks, and neither do Filter or Paths steps, failed or halted actions, or several utility steps such as Formatter, Delay, Looping and Storage. Only successful action steps count, which is why a three-step Zap often uses two tasks.

Why did my Make scenario use so many operations?

Bundles multiply. A trigger module runs once no matter how many bundles it returns, but every module after it runs once per bundle. A trigger returning ten records through a five-module scenario is around fifty operations, not five. Filtering early is the main way to control this.

Is n8n cheaper than Zapier?

It depends entirely on workflow shape. n8n's per-execution model makes complex, many-step workflows and big loops cheap, so it wins decisively there. A very high-frequency workflow with only one or two actions can favour per-task pricing. Model your own volume against each vendor's counting rules.

How do I reduce my automation bill?

Design to the unit you pay for. On per-action or per-operation platforms, filter early so cheap steps discard work before expensive ones, batch records instead of looping one at a time, and consolidate steps. On per-execution platforms, batch the trigger so one run handles many records.

Do AI steps count as extra?

Twice over. The AI step consumes the platform's billing unit like any other, and separately you pay the model provider for tokens. Those are different bills from different vendors, and forgetting the second one is a common way for an automation budget to come in low.

Primary source

Make - how operations are counted

Checked August 2026. Anything priced or rate-limited moves, so verify against the source before you budget.

Want this built rather than explained?

Knowing what execution vs task vs operation means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.