Not every repetitive task deserves software. Here's the filter we use in the first call to find the one that pays for itself fastest.
Most teams pick the workflow that annoys them most. That's rarely the one that pays back fastest. The right first automation is high-volume, rule-heavy, and low-judgement — something your team repeats weekly and could describe step by step without hesitating.
We score candidate workflows on four things: how many times a week it runs, how long each run takes, how often a mistake costs real money, and how clearly the rules can be written down. A task that scores high on all four is a build. A task that scores high on frustration alone is usually a process problem, not a software problem.
The output of that session is a one-page definition of done: the inputs, the rules, the systems it touches, and what the finished output has to look like. Everything after that is engineering.