Automate for outcomes

Effective process automation is essential for modern business. Handing mundane tasks off to a computer frees up people to solve the difficult problems. However, poorly designed process automation can cost more people-time than having no automation at all.

The collaborate-build-learn loop

Designing process automations is a lot like designing software systems. It consists of four steps, three of which occur in a loop.

• Collaborate with the process owners to fully understand the process
• Iterate on:
  • Decide which individual steps to automate in collaboration with the process owners
  • Spend at most one week implementing what you can
  • Give it to the process owners and listen to their feedback

As with any kind of design, listening to your users is absolutely essential.

Two rules for process automation design

The high level steps of automating a process are not difficult to understand, but they don't tell the whole story. Avoid common pitfalls by doing these two things right.

Automate the whole function

Why Doormen Are Not Replaced With Automatic Doors: define the true function before you automate. By defining the function of a doorman as "Someone who opens doors" this can be automated using a sliding door and an infrared sensor. However, a doorman does much more than just opening doors. When automating a process, take into account all aspects of that process.

Avoid automating decisions

Computers are not good at making decisions. Instead: give the human user pertinent data to base their decision on, plus a way to act on that decision immediately. This is much easier (faster, cheaper) to implement and keeps human users in the loop.

If at some point you receive feedback along the lines of "If we tighten up these rules the machine can take the decision for me", this is now easy to do and happens in full agreement with your users.

Assist, don't replace

The goal when automating any business process is to strike the right balance. Move work from humans to computers provided that the work suits computers. If something is time-consuming or complicated for a human, find areas where computers can assist. Expecting a complex process to be fully automated often results in time wasted by both the developers and the users.

A small, accurate, and robust process automation does more than just save time. It instils users with confidence that this process is taken care of automatically and frees up their mental bandwidth to focus on doing their own best work.