Automation ROI: how to check it pays off before you build

Start with the manual cost. Time per task, times tasks per month, times the cost of an hour of that person's time. Use the fully loaded cost (salary plus the extras an employer pays), not just the salary figure.

Then add the cost of mistakes: how many tasks need correcting, how long each correction takes, and whether any error reaches a customer or a supplier. Now you have a monthly figure for doing it by hand.

The automation isn't free to run, so subtract its monthly cost: hosting, software fees, usage charges and a little maintenance time. What remains is the net monthly saving. Setup cost divided by net monthly saving gives you the payback period in months.

These numbers are an example, not a quote and not a benchmark. Say your team enters supplier invoices by hand. Replace each figure with your own and the method still holds.

Say 400 invoices arrive each month and each takes 6 minutes to enter. A person costs $15 an hour, fully loaded. After automation, 10% of invoices still need a person to review them.

Not the most annoying one. The best first candidate is boring, frequent and rule based. If the same task happens hundreds of times a month and follows steps a new hire could learn in a day, the calculation above usually works.

Tasks that need judgment on every case, or that change rules every few weeks, are poor first choices. They cost more to build and more to maintain.

The usual mistake is counting what the automation saves and not what it needs. Some of these costs are small. Some are the difference between 8 months and 18.

Don't forget the savings side either: time spent chasing errors that no longer happen, and the days a task used to wait because the one person who did it was away.

If payback comes out at three years, or the saving depends on volume you hope to reach someday, wait. Put the money elsewhere, or do a cheaper version: a template, a checklist or a simple spreadsheet formula.

Be careful with the hours-saved argument too. Saving 36 hours a month only counts as money if those hours are used on something useful or let you avoid a hire. Otherwise it's a nicer Monday, which is fine, but a different kind of return.