Automation Pricing Models Compared: How You Actually Get Billed

Every automation platform bills on a different unit, and understanding automation pricing models matters more than the sticker price. One tool charges per task, another per operation, another per active flow, another per second of compute. The same workflow can cost wildly different amounts depending on which unit you’re being metered on, and if you resell automation to clients, that unit is what decides your margin.

Task, operation, credit, transaction, active flow, compute-second: the words get used loosely and the differences are real money. This is a plain-language breakdown of the main automation pricing models, the same workflow costed across each, and why the billing unit deserves more of your attention than the headline price. I build one of these platforms, so I’ll be upfront about where each model wins, including the ones that aren’t mine.

The main ways automation platforms bill you

There are five billing models in common use. Each answers a different question about what you’re paying for, and the differences show up fast once your workflows get complex or your data volume grows.

• Per task or per step. You pay for each action a workflow performs. A ten-step workflow costs roughly ten units per run. Used by Zapier (tasks) and Albato (transactions).
• Per credit, scaled by data. You pay a credit for each module, on each record it processes, so cost multiplies with volume. Used by Make.
• Per active flow. You pay a flat monthly fee for each enabled workflow, with unlimited runs. Used by Activepieces.
• Per compute credit. You pay by how long a run takes, often with AI billed separately on top. Used by Latenode.
• Per compute time. You pay for the actual runtime a workflow uses, regardless of step count or record count. Used by TaskJuice.

Per-task and per-transaction: you pay for every step

On the per-step models, each action a workflow takes is a billable unit. A workflow with ten action steps costs about ten units every time it runs. Triggers are usually free, but the work is not, and processing a batch of records runs each step against each record.[1][5]

This is the Zapier task model and the Albato transaction model. It’s simple to understand and fine for light workflows, but it charges you for complexity: every step you add to make a workflow smarter adds to the bill on every run. I broke down the agency impact in the Zapier task tax and the Albato comparison.

Per-credit: you pay for every step, times your data

Make’s credit model is per-step billing with a multiplier. Each module bills a credit for every record it processes, so a scenario that runs nine steps over 100 records consumes on the order of 900 credits in a single run. Routers and filters are free, but the modules doing the work are not, and even testing a scenario consumes credits.[2]

The multiplier is the part that surprises people. Your cost grows with your client’s data volume, which is exactly the thing you can’t control and don’t want tied to your bill. The full breakdown is in the Make comparison.

Per-active-flow: you pay per workflow, not per run

Activepieces flips the model. You pay a flat monthly fee per active flow, around $5 after a free allowance, with unlimited runs on each. A flow that fires a million times costs the same as one that fires once.[3]

This is genuinely the cheapest model for a single high-volume workflow, and I’ll give it that. The catch for an agency is the opposite shape: you accumulate many low-volume workflows across a book of clients, and you pay for every enabled one whether it runs or not. It charges you for breadth of automation, which is exactly what an agency builds. More on that in the Activepieces comparison.

Compute-credit and compute-time: you pay for time, not steps

The last two models bill by how long a workflow actually runs. Latenode uses compute credits, where one credit buys roughly 30 seconds of run time, though its managed AI is billed separately as tokens on top.[4] TaskJuice bills straight compute time, the runtime a workflow uses, with a monthly allowance built into each plan and AI run on your own provider key so there’s nothing stacked on top.

The advantage of time-based billing is that step count and record count stop being pricing dimensions. A ten-step workflow and a three-step workflow that finish in the same time cost the same, and a batch of 100 records is one run rather than 100 multiplied by your step count. The compute-credit details are in the Latenode comparison.

The same workflow, costed five ways

Take one workflow: a trigger plus nine action steps, processing 100 records in a run. Here’s roughly how each model bills that single run, holding the work constant.

• Per task or per transaction: about 900 units, nine steps across 100 records.
• Per credit: about 900 credits, same multiplication, plus extra for any AI step.
• Per active flow: flat, it’s one enabled flow, so the run itself adds nothing.
• Per compute credit or compute time: the seconds the run takes, so 100 records in one pass costs the run’s runtime, not a multiple of it.

Three of the five models multiply with your steps and your records. One is flat per workflow. One tracks time. That’s the whole game. The right question isn’t which platform has the lowest headline price, it’s which unit matches how your workflows actually behave.

Which model protects an agency’s margin?

If you resell automation to clients at a fixed monthly retainer, the billing model decides whether that retainer holds. Per-step and per-credit models tie your cost to your client’s complexity and data volume, so your margin shrinks every time their business grows. That’s a bad trade when you’ve already quoted a flat price.

Per-active-flow is predictable but charges for breadth, so a growing book of client workflows becomes a growing base fee. Compute-time keeps step count and record count out of the equation entirely, so a workflow getting smarter or a client’s data growing doesn’t automatically move your cost. For an agency that lives on predictable margins across many client workflows, that predictability is the point. The unpredictable-bill problem is real: usage-based platforms produce surprise invoices when volume spikes, and a fixed retainer has no room to absorb one.

Frequently asked questions

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

A task or operation is usually one action step, so a multi-step workflow uses several per run. A credit is the same idea but billed per record as well. An active flow is a whole enabled workflow billed monthly regardless of runs. An execution or compute-second measures the run itself, not the steps inside it.

Is per-task or per-operation pricing cheaper?

It depends entirely on your workflows. Per-task and per-credit models get expensive fast with many steps or high data volume, since both multiply. Flat-per-flow and compute-time models stay cheaper as complexity and volume grow, because neither charges per step. For light, simple workflows the per-step models can be perfectly cheap.

Do triggers count as tasks or operations?

Usually not on the per-step models, triggers are typically free and only action steps are billed. But per-credit platforms count each polling check as a use, so a trigger that checks for new data every few minutes can quietly consume credits even when nothing changes.[2]

Why is my automation bill higher than expected?

Almost always because your data volume grew and your model multiplies. On per-step and per-credit billing, more records means more billable units, and testing and failed runs can count too. Overage rates are often higher than in-plan rates, so the worst bill arrives exactly when you’re busiest.

Which automation pricing model is best for agencies?

For agencies reselling at fixed retainers, compute-time billing tends to protect margin best, because it doesn’t multiply with the steps or data volume you can’t control. Per-active-flow is a close second for predictability. Per-task and per-credit are the riskiest to build a resale business on, since your cost rises with your client’s success.

There’s no universally cheapest model, only the one that fits how your workflows behave. Light and simple, per-step is fine. One huge high-volume flow, per-active-flow wins. Complex, variable, resold to clients at a fixed price, compute-time keeps your margin off the roller coaster. TaskJuice bills compute time on purpose, because it’s the model that lets an agency quote a client and keep the difference. TaskJuice is opening access to founding agencies now. Join the early-access list to see how the math works for your book.

References

[1] Zapier: How Pay-Per-Task Billing Works: help.zapier.com/hc/en-us/articles/15279018245901-How-pay-per-task-billing-works-in-Zapier

[2] Make Help: How Features Use Credits: help.make.com/how-features-use-credits

[3] Activepieces Pricing: activepieces.com/pricing

[4] Latenode Billing System: Credits and Plug-n-Play Tokens: documentation.latenode.com/account-management/billing-page-and-managing-subscription

[5] Albato Help: Steps and Transactions: albato.com/wiki/articles/steps-per-automations