Automation Pricing: Tasks vs Credits vs Seats
Zapier bills successful actions, Make bills every module run including trigger checks, and n8n bills whole workflow runs. The same six-step order workflow at 1,000 orders a month burns 4,000 units on one meter and 7,880 on another, so comparing headline prices alone tells you almost nothing.
The price on an automation platform's pricing page is almost never the price you pay. What determines your bill is the billing unit — the meter — and every major vendor uses a different one. The same workflow, moving the same data, can consume 1,000 units on one platform and 7,880 on another. Before you compare dollar figures, you have to compare meters.
Here is what each meter actually counts, verified against the vendors' own documentation in September 2026, and a worked example showing how far apart the same job lands.
The six meters you are actually buying
1. Tasks (Zapier)
Zapier bills per successful action. Its documentation is unusually specific: "A task is any successful action that runs in Zapier. Only successful actions count toward your task usage." Triggers are free. So are Filter, Paths, Formatter, Delay, Looping, Digest, Storage, Zapier Manager, Tables and Forms steps. Actions that error or halt do not count, and neither do steps skipped because a filter stopped the run (Zapier, How is task usage measured, accessed September 2026).
The practical consequence: on Zapier, a six-step workflow may bill as four tasks. Data-shaping is free; touching an outside service is not.
2. Credits, formerly operations (Make)
Make counts every module run. Its pricing page states: "Each module action in your scenario, like adding a Google Sheet row or fetching Gmail account data, counts as one credit" (Make pricing, verified September 2026). Make renamed operations to credits in an announcement dated January 19, 2026, with the note that "Your credit usage won't increase—1 operation still uses 1 credit."
The part that catches people is triggers. Make's documentation describes trigger modules as consuming one operation per check — the example given is "the google sheet > watch new rows module checks for new rows in a google spreadsheet = 1 operation (one per check)" (Make, Operations, accessed September 2026). A scenario polling every 15 minutes checks roughly 2,880 times a month whether or not anything happens. That is the opposite of Zapier's rule, and it is the single biggest source of surprise on Make bills.
3. Executions (n8n)
n8n counts whole workflow runs. Its pricing page defines an execution as "a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes. It's still a single execution" (n8n pricing, verified September 2026). Plans are sold in execution counts — Starter at €20/month billed annually includes 2,500 executions with unlimited steps; Pro at €50/month includes 10,000.
Under this meter, workflow complexity is free and workflow frequency is everything. A 40-step monster costs the same as a two-step one.
4. Seats (Power Automate, Airtable)
Microsoft licenses Power Automate per user or per automation. Power Automate Premium is $15.00 per user per month, paid yearly; Power Automate Process is $150.00 per bot per month; Power Automate Hosted Process is $215.00 per bot per month (Microsoft Power Automate pricing, verified September 2026). Microsoft's own documentation frames the split plainly: a user license is "user-centric," while a capacity license is "automation-centric" and is "allocated to a Power Automate automation" (Microsoft Learn, Power Automate licensing types).
Airtable is per-seat too — $20 per user per month on Team and $45 per user per month on Business, both billed annually (Airtable pricing, verified September 2026). Seat meters are the easiest to forecast and the easiest to blow up: growth in headcount, not growth in volume, is what moves the bill.
5. Premium connectors
On Microsoft's platform, the connector itself is a licensing gate. The documentation is explicit: "The premium connectors entitlement is needed to add a premium connector to a cloud flow and trigger it" (Microsoft Learn, Power Automate licensing types). That entitlement comes with Power Automate Premium, Process or Hosted Process — not with the free trial or a seeded Microsoft 365 license. If the one system you need to reach sits behind a premium connector, your floor is a paid license regardless of volume.
6. Platform request limits
This is the meter almost nobody reads, because it does not appear on the pricing page at all. Microsoft caps Power Platform requests per user per 24 hours: 40,000 for Power Automate Premium, 250,000 per license for Power Automate Process and Hosted Process, and 6,000 for Power Automate Free and Office 365 seeded licenses (Microsoft Learn, request limits and allocations, accessed September 2026). You can be well inside your license count and still hit a wall.
The meters disagree on a basic question: does checking for work count as work? Zapier says no — its pricing page states it "never charges you a task to check for new data." Make says yes — one operation per trigger check. That single difference can double a bill.
The same workflow, four meters
Take a common small-business flow: a new order arrives, the workflow looks up the customer in a CRM, formats the shipping address, writes a row to a spreadsheet, posts to Slack, and sends a confirmation email. That is one trigger plus five actions, one of which is pure data formatting. Assume 1,000 orders a month and a 15-minute polling interval where the platform polls.
| Platform | What it counts | Monthly units for this workflow | Entry plan that covers it |
|---|---|---|---|
| Zapier | Successful actions only; trigger and Formatter are free | 4 actions × 1,000 = 4,000 tasks | Above the 750-task entry tier; task tier must be raised (Professional starts at $19.99/mo) |
| Make | Every module run, including each trigger check | (2,880 polls) + (5 modules × 1,000) = 7,880 credits | Core, $12/mo for 10,000 credits |
| n8n | Whole workflow runs, unlimited steps | 1 × 1,000 = 1,000 executions | Starter, €20/mo annually for 2,500 executions |
| Power Automate | Per-user license, plus a daily platform request cap | 1,000 flow runs, billed by seat not by run | Premium, $15.00/user/mo paid yearly (premium connectors included) |
All plan figures above were checked against the vendors' pricing pages in September 2026. One reporting note: Zapier's pricing page presents Professional task tiers in a selector rather than as a published list of per-tier prices, so we can confirm the $19.99 entry point at 750 tasks but not the price at 4,000 tasks without configuring a cart.
Unit counts for the identical job range from 1,000 to 7,880. Comparing headline prices without comparing meters tells you nothing.
How the meter changes the way you should build
Once you know which meter you are on, some build decisions become obvious.
- On a per-action meter, push filtering as early as possible. A filter costs nothing and every action it prevents is money. Consolidate three small updates into one.
- On a per-module meter, attack polling first. Switching a 15-minute poll to an instant webhook removes roughly 2,880 monthly checks before you optimize anything else.
- On a per-execution meter, batch. One workflow run handling 50 records costs the same as one handling one. Scheduled batch jobs beat per-record triggers.
- On a seat meter, consolidate ownership. Volume is nearly free; the cost is how many people need to touch the thing.
Questions worth asking before you sign
- Does a trigger check consume a unit when nothing has changed?
- Do failed or retried steps consume units?
- Which built-in steps are free, and are AI steps among them?
- What happens at the limit — does the vendor hold runs, or bill overage automatically? Zapier's documentation says paid plans with pay-per-task billing are charged for overage, while plans without it have runs held "until your billing cycle resets."
- Is there a second, undocumented-on-the-pricing-page cap, like a daily API request ceiling?
- Do the connectors you specifically need require a higher tier?
Bottom line
There is no universally cheap platform. There is a platform whose meter happens to match the shape of your work. High-frequency, low-complexity work is punished by per-action meters and rewarded by per-execution meters. Complex, low-frequency work is the reverse. Heavy polling is punished by Make's meter and free on Zapier's.
Count your units before you compare prices. Sketch the workflow, count the steps, estimate the monthly volume, then apply each vendor's stated rule. It takes twenty minutes and it is the only comparison that means anything.