Two automation tools advertise "2,500 included" on their entry plan. One is a far better deal than the other — and the pricing page will never tell you which. The reason is buried in a single word that every platform defines differently: the billing unit. Zapier sells you tasks, Make sells operations, n8n sells executions, Power Automate sells Power Platform requests, and Pipedream sells compute credits. They sound interchangeable. They are not. The same workflow, run the same number of times, can burn 5,000 units on one platform and 1,000 on another.
Why the included number on a pricing page is a trap
Pricing pages are designed to be scanned, not decoded. Every vendor publishes a headline allowance — "750 tasks," "10,000 operations," "2,500 executions" — and trusts that you will read those as roughly equivalent quantities. They are not equivalent, because each number counts a different event. The unit is the catch, not the price.
Here is what each platform actually meters, verified against current vendor pricing and documentation:
| Platform | Unit | What consumes one unit | Steps multiply? |
|---|---|---|---|
| Zapier | Task | Each successful action step | Yes |
| Make | Operation / credit | Every module that runs, including the trigger | Yes |
| n8n | Execution | One full workflow run, any number of nodes | No |
| Power Automate | Power Platform request | Every API call: triggers, actions, retries, loops | Yes |
| Pipedream | Credit | Each 30 seconds of compute time | No (time, not steps) |
| IFTTT | Applet run | Effectively unlimited; you pay per applet, not per run | No |
Read that "Steps multiply?" column twice. It is the whole game. On Zapier, Make, Power Automate, and Workato, a longer workflow costs more every single time it fires. On n8n and Pipedream, workflow length is nearly free — you pay per run (n8n) or per second of runtime (Pipedream), not per box on the canvas.
The three metering philosophies
Under the marketing vocabulary, there are really only three ways to count.
Per-action: you pay for every step
Zapier, Make, Power Automate, and Workato all meter per step. On Zapier's per-task model, only successful action steps count — triggers, Filters, Paths, and Formatters are free — so a 4-action Zap costs 4 tasks per run. Make counts more aggressively: every module, including the trigger and any filters or routers, consumes an operation (Make renamed "operations" to "credits" in August 2025, at a 1:1 conversion). Power Automate is the most literal of all — every API call counts, including internal steps like initializing a variable, plus retries and each iteration of a loop.
The implication: on per-action platforms, the cost of a workflow scales with its complexity. Add an enrichment step, a second branch, an extra notification — each one multiplies against your run volume.
Per-run: complexity is free
n8n counts one execution per full workflow run, regardless of whether the workflow has two nodes or fifty. A 50-step n8n workflow and a 2-step n8n workflow cost exactly the same per run. This is a structurally different deal for complex automations — and it gets more extreme if you self-host the n8n Community Edition, where executions are unlimited and your only cost is the server. (Note: n8n removed its free cloud tier in late 2025, so the free path now runs through self-hosting.)
Per-compute: you pay for time
Pipedream ignores steps entirely and bills compute credits — one credit per 30 seconds of execution at the default memory allocation. A 10-step workflow that finishes in 200 milliseconds costs the same single credit as a 1-step workflow that finishes in 200 milliseconds. The variable that drives your bill is runtime (and memory), not how many boxes you wired together.
How to compare them: convert to cost per workflow run
The only honest comparison is to pick a real workflow, count how many units each platform would charge for it, and divide by what that allowance costs. Take a common shape: a 5-step automation — one trigger plus four actions — that runs 1,000 times a month.
Same workflow, 1,000 runs/month — units consumed
- Zapier: 4 actions × 1,000 = 4,000 tasks (trigger is free). The $19.99 Professional plan includes 750 tasks — you would need a much larger tier.
- Make: all 5 modules × 1,000 = 5,000 operations. The ~$21 Core plan's 10,000 operations covers it comfortably.
- n8n: 1 execution × 1,000 = 1,000 executions. The ~$21.70 Starter plan's 2,500 executions covers it with room to spare — or unlimited if self-hosted.
- Power Automate: every step plus internal actions × 1,000 = 5,000+ requests drawn from your per-user daily request pool.
- Pipedream: if each run finishes under 30 seconds, 1,000 credits.
Same automation, same volume: 4,000 units on Zapier, 5,000 on Make, 1,000 on n8n, 1,000-ish on Pipedream. The "included" headline numbers on the pricing pages — 750, 10,000, 2,500 — told you almost nothing about which platform is cheapest for this workflow. The shape of your automations decides the winner.
Where enterprise platforms hide the unit entirely
Workato, MuleSoft, Boomi, Tray.io, and Celigo mostly do not publish unit allowances at all — pricing is custom-quoted, and the billing unit is one variable inside a larger negotiation that also includes connector tiers, environment fees, and platform licenses. Workato, for example, uses a task-based usage fee on top of a base platform fee, but the per-task definition is buried in the contract. If you are evaluating that tier, the unit math here still applies — it is just one line in a bigger quote. We break down how to read those quotes in enterprise iPaaS pricing decoded.
What to do before you sign up
Do not pick a platform by its included allowance. Pick a representative workflow, count its billable units on each platform using the table above, and project that against your real monthly run volume. If you are already on a per-action platform and your workflows have grown long, that unit math is often the single biggest argument for switching — we ran the full cost of leaving Zapier for exactly that reason. The cheapest sticker price and the cheapest workflow are frequently two different tools.
Compare all 18 platforms on real terms
See pricing, free tiers, integration counts, and feature depth side by side — normalized, not marketing-speak.
Compare automation platforms →Frequently asked questions
Do triggers count as tasks on Zapier?
No. Zapier only charges a task for each successful action step. The trigger that starts the Zap, along with built-in Filters, Paths, and Formatters, does not consume tasks. Failed action steps also do not count.
Why does Make burn through its allowance faster than it looks?
Because Make counts the trigger and every internal module, not just the actions that change data. A scenario you think of as "three steps" may be metered as four or five operations per run once you include the trigger and any filters or routers.
Is n8n really unlimited if I self-host it?
The self-hosted Community Edition places no cap on executions — you pay only for the server it runs on. The trade-off is that you own setup, maintenance, and uptime. We compare that total cost against managed cloud plans in self-hosted vs cloud workflow automation.
What is the single best way to compare two platforms' pricing?
Convert both to cost per run of one representative workflow. Count how many billable units that workflow consumes on each platform, divide the plan price by the included allowance to get a per-unit cost, then multiply by your monthly run volume. That number — not the headline allowance — is the comparison that matters.