How Automation Work Gets Priced, and How Not to Get Burned Buying It

Full disclosure before we start: we sell automation builds. Which makes this either a shameless conflict of interest or a useful insider's guide, depending entirely on how it reads to you. I'm going for the second.

Whether you're buying automation work from a freelancer, an agency, or from us, you'll run into three pricing models. Each one quietly tells you something about the person on the other side of the invoice.

Fixed project price

"$X to build your invoice pipeline." Clean, easy to budget, and it forces the builder to scope the job properly before a line gets written. The risk hides in the change orders. A builder who lowballs the fixed number is often planning to make it back the moment reality turns up and the spec "changes."

What to check. Whether the price includes a proper hand-over. Documentation, credentials sitting in your accounts, a walkthrough you can actually follow. A build you can't run without phoning the builder every week isn't an asset. It's a subscription wearing a project's clothes.

Retainer

"$X a month for automation support." Sometimes entirely legitimate, because systems drift and APIs break without asking permission first. But a retainer offered before a single thing has been built is a yellow flag worth staring at. The incentive quietly rotates from "get this finished" to "keep this needing me."

What to check. What actually happens each month, spelled out in specifics rather than slogans. "Monitoring and maintenance" should arrive with logs you're allowed to look at. Our own bias is to build it, hand it over, and only charge again when there's genuinely new work. We wrote "Built to Hand Over" into our values precisely because the opposite drives us mad as buyers.

Per-execution pass-through

Increasingly common with n8n, where the paid tiers bill per workflow execution. The builder passes the infrastructure cost through to you, usually with a bit of margin on top. Perfectly fair, but ask for the unit maths in writing. A workflow that costs pennies at today's volume can hand you a nasty surprise at 10x, and if it's any good, you will hit 10x.

The one question that sorts everyone

"Walk me through what happens when this breaks at 2am." Good builders have a real answer, involving error handling, alerts, and a fallback path, because good builders have all been paged by their own work at some ugly hour. Bad builders tell you it won't break.

It will break. Everything breaks eventually. The pricing model matters far less than whether the person taking your money has genuinely planned for the night it does.