Workflow automation: which processes are worth it first

01The four criteria for good candidates

Workflow automation means a process that used to run by hand now runs on its own. Not every process is suited for it, and the selection decides between success and frustration. Four criteria have proven themselves in my projects:

• Frequent: the task comes up daily or weekly, not twice a year. Frequency is the lever that turns small minutes into large sums.
• Rule-based: the steps can be written down so precisely that a stranger could follow them without asking questions all the time.
• Error-prone by hand: transposed digits, forgotten mails, customers created twice. Wherever people retype data, mistakes happen, and diligence does not change that.
• Measurable: you can say how much time the task costs today and how you will recognise the improvement afterwards.

If all four apply, the case is almost always clear. If only one applies, there are very likely better candidates in your daily work.

02What manual work really costs

A calculation, deliberately conservative:

• Transferring order data by hand takes 30 minutes per day.
• Across 220 working days that is 110 hours per year.
• At an internal rate of 40 euros per hour: 4,400 euros per year, every year again.
• Not included: fixing errors, answering queries and the tasks left undone in the meantime.

Run this with your own numbers: minutes per day times hourly rate times working days. The yearly cost of manual work is the most honest argument for or against an automation project, in both directions.

03Typical first projects

First automations look similar across small businesses, and that is no coincidence: these tasks meet all four criteria, are manageable in scope and show visible results quickly.

• Transferring order data from the shop into inventory or accounting.
• Reading incoming invoices automatically, naming them and filing them in the right folder.
• Generating recurring reports, for example weekly revenue, open items or stock coverage.
• Notifications when something needs your attention: low stock, a missing payment, an unusual order.

04Why starting small beats planning big

The reflex in many businesses is to tackle every process at once. My experience says otherwise: one small process that runs after two weeks convinces more than a concept for twelve processes that is still not live after half a year. The small win also shows you how your team handles automation, where trust is missing and which edge cases really exist. You learn more from that than from any planning round, and the next project gets better because of it.

05When automation is not worth it

• Rare edge cases: what happens twice a year is done faster by hand than you can maintain the automation.
• Unstable processes: if the workflow changes every few weeks, you rebuild the automation more often than it saves you time.
• True judgement calls: if every case needs a human decision, software can prepare the decision but not make it.

Automation does not repair a broken process. Clean up the workflow first, then automate, otherwise the same mistake just produces results faster.