Automation promises quick wins: less manual entry, shorter cycle times, fewer errors. Applied to a process nobody has analysed, however, it tends to lock in the defects.
What gets automated by mistake
- redundant checks created to compensate for unreliable upstream data;
- approval steps that add no information;
- data transfers between systems that could be integrated;
- exceptions so frequent they are effectively the rule.
Understand first, then automate
Reconstructing the real process — not the one described in the procedures — reveals waiting, rework and error causes. Part of the work can often be eliminated; another part simplified; only what remains is worth automating.
Handling exceptions
A good automated system does not hide exceptions: it surfaces them, classifies them and routes them to the right person with the information needed to decide. This is where the difference between speed and reliability is measured.
The right question is not “what can we automate?” but “which result must improve, and which work is really needed to achieve it?”.
