This is the first thing I do in a paid audit, reduced to two pages you can run yourself. Score each process out of five on four measures, and the ones worth building announce themselves.
It will also tell you which processes to leave alone — which is usually the more expensive mistake to get wrong.
- 01
Volume
How many times a week does this actually happen? Below a threshold, the build costs more than the hours it saves.
- 02
Variability
Is every instance broadly the same, or is each one a judgement call? Sameness is what makes automation hold.
- 03
Rule clarity
Could you write the rules down for a new starter? If nobody can, that is the finding — and it is worth having.
- 04
Error cost
What happens when the output is wrong? High consequence does not rule automation out, but it changes what you build.