Start with the process, not the software

Write down the steps of the task as they really happen today, including the small ones: who copies what, who waits for whom, where something gets checked twice. Very often a step exists only because of an old workaround that nobody needs any more.

Removing a step costs nothing. Automating it costs money and needs looking after. So the first question is always whether the step should exist at all.

Four questions before automating anything

  • How often does it happen? A task done twenty times a week is a good candidate. One done twice a month rarely is.
  • Is it the same every time? Software follows rules, so it needs a pattern. If every case needs judgement, a person will still be doing most of the work.
  • What does a mistake cost? A reminder sent at the wrong time is a small embarrassment. A wrong invoice or a wrong answer to a health question is not. The higher the cost, the more checks, and the more a person stays involved.
  • Does a tool you already pay for do it? Booking systems, accounting software and email tools often have reminders, templates and rules that were simply never switched on.

Simplify first when

  • Each person does the task differently. Agree on one way first, then automate that way.
  • Information arrives in five formats: by phone, text, email and on paper. One short form often solves more than any automation.
  • Things wait for approvals from people who only pass them on. Remove the step instead of speeding it up.

Automate when

  • The steps are stable and repeat often.
  • Delay costs you something, like an inquiry that arrives at 9 pm and waits until morning.
  • People copy the same details from one program to another.
  • Someone has to remember to do it, and sometimes doesn't.

A quick way to tell if it's worth it

Multiply the hours a week the task takes by what an hour of that person's time is worth, then by 52. That's roughly what the task costs you in a year. Compare it with the fixed price of automating it. The number leaves out the harder-to-count costs, such as mistakes, slow replies and the client who went elsewhere, so treat it as a floor, not a promise.

Leave the exceptions to people

The best automation handles the routine cases and hands the unusual ones to a person, with the details attached. Trying to automate every exception makes the system fragile and expensive. And whatever it does, a failed step should be reported to someone, so nothing gets lost quietly.

If you're not sure which of your tasks fall on which side, that's exactly what the free call is for. Sometimes the answer is a better process and no software at all, and I'll say so.