Three questions to ask before you automate anything
Most failed automations were doomed at the choosing stage, not the building stage. These three questions take ten minutes and catch nearly all of them.
The most expensive automation mistakes are not technical. They are choosing the wrong process, and you can catch almost all of them before writing anything by answering three questions honestly.
One: does this process happen the same way every time?
Not roughly the same way. The same way.
If the answer involves « it depends » more than once, you have not found an automation candidate; you have found a process that needs to be described before it can be anything. Describing it is valuable work in itself, and many companies discover during that exercise that two departments have been doing it differently for years.
The 28 January post on writing the process down first still applies, and it is still the step most often skipped.
Two: what happens when it gets it wrong?
Every automation will produce a wrong output eventually. The question is what that costs.
A mis-summarised internal email costs nothing; somebody reads the original. A mis-filed supplier invoice costs an afternoon. A wrong figure sent to a customer costs a relationship. Sort your candidates by this, not by how much time they would save, and automate upward from the cheap failures.
Three: who notices?
The dangerous automations are not the ones that fail loudly. They are the ones that fail quietly and keep running for four months.
Before you build, name the person who would notice, and how. If the honest answer is « nobody would notice », either build the check first or do not build the automation.
The order these should be asked in
Three first. It is the one most often skipped and the one that turns a saving into a liability.
A process that passes all three is worth automating even if it saves only twenty minutes a week. One that fails question three is worth avoiding even if it would save a day.