Business Technology
How to identify a process worth automating
Not every slow process should become software. The ones that should are frequent, expensive, and already somewhat structured.
A process is a candidate when it happens often, follows a recognizable pattern, and costs real time or error when done by hand. A rare, political, or highly judgment-based task is usually the wrong first target.
Start with frequency and duration. A 20-minute task done twice a year is a nuisance. A 20-minute task done forty times a week is an operating cost. Then ask whether the inputs are available in a system, a file, or a person’s memory. Memory does not automate well.
Look at exceptions. If every case is special, you do not have a process yet — you have improvisation. Software can still help, but the first step is to name a standard path and a short list of exceptions, not to encode every special case.
Finally, check who owns the outcome. Automation without an owner becomes a silent failure. Someone must notice when the output is wrong, and someone must be trained to use the new path.
If you can answer those questions, you are ready for a serious conversation — audit, design, or a small first version. If you cannot, the useful work is still upstream: clarify the process before choosing a tool.