The question almost always arrives the same way: "we have a process that breaks every week — what would it cost to automate it?" And the honest answer isn't a figure, it's a range with conditions. This article explains where that range comes from.

What you pay once and what you pay monthly

A complete internal workflow — intake form, approval, notifications and traceability — starts at €1,500 to implement plus a retainer from €250 a month. Two separate lines, because they answer two separate things.

Implementation covers mapping the real process, building the workflow, testing it in an isolated environment and documenting it. The retainer covers what happens afterwards: incidents, permission changes, connectors Microsoft updates, small adjustments that under any other model would be a fresh quote.

A diagnosis without implementation starts at €450, is a one-off payment, and is deducted if we work together afterwards. That's what I recommend when it isn't yet clear what's worth automating.

€1,500to implement, paid once
€250a month for the retainer, and it's optional
€450for the diagnosis on its own, deducted later

The four things that move the price

How many hands touch the process. A workflow with one requester and one approver is one problem; one with three departments, an external supplier and an exception for the board is another. Every person adds a state, and every state adds a failure path that has to be tested.

What already exists, and the shape it's in. Finding an application that works with its documentation up to date makes the work cheaper. Finding three lists with nearly identical names, permissions inherited from a personal account and a flow nobody dares touch makes it more expensive — and that's the more common case.

Whether a test environment is needed. When the process is in production and can't be stopped, a parallel site and parallel lists have to exist before anything is touched. That's real work that doesn't show up in the result, and it isn't negotiable: no change should make its debut with real orders inside it.

How many exceptions have to be supported. "Except in August", "except when the order comes from the branch office", "except when Marta asks". Every exception is a branch, and it's worth deciding which ones get automated and which stay a human decision.

Automating every exception can cost more than the main process.

What you usually don't have to buy

Almost no company needs new licences to start. Most already pay for Microsoft 365, and that includes Power Automate, Power Apps and SharePoint — which is where the workflow gets built. When a process doesn't fit there — external integrations, long-running tasks, orchestration — I use self-hosted n8n.

A new licence only gets proposed when there's no reasonable way around it, and it's said before we start, not on the invoice.

The comparison that matters isn't with zero

Budgets tend to get compared against doing nothing, and that's the wrong comparison. The right one is against what the current process costs: the hours of the person forwarding emails, the order that gets lost and redone, the afternoon someone spends reconstructing the state of everything, and the risk of the whole chain depending on one individual account.

When that calculation doesn't favour automating, I say so.

Some processes are fixed by sorting out who does what, without touching a single tool, and charging to automate those would be selling smoke.

How to find your number

There's no way to give a figure by email without seeing the process. What can be done in a twenty-minute call is place it in a range and tell you whether it's worth going further.

If you have a process that breaks every week, tell me in four lines: how it works today, which part breaks, and who suffers it.