A practical framework for evaluating business automation opportunities

Most automation backlogs are wish-lists, not portfolios. They are ranked by who shouted loudest in the last QBR rather than by economic logic. Within a quarter, the loudest items are half-built and the highest-value ones are still parked.

Score every candidate workflow on four axes. Frequency: how often does this fire? Cost per occurrence: labour, error remediation, downstream impact. Process clarity: is the current workflow stable enough to automate, or does it need redesign first? Reversibility: how easily can a failure be detected and rolled back?

High frequency + high cost + high clarity + high reversibility is the quadrant you build in first. High frequency but low clarity is a process redesign opportunity disguised as an automation one. Low frequency but high cost might be a checklist problem, not an automation one.

The single most expensive mistake in this space is automating a broken process at high fidelity — you now have a faster, more reliable bad process. Always ask: if we did nothing else, what would the right version of this workflow look like? Then automate that one.

The TCO of an automation includes the cost of someone fixing it at 11pm in two years. Platforms your team understands, can extend, and can debug beat clever platforms they cannot — every time.

Treat every automation as an experiment with a measurable hypothesis: time saved, error rate, throughput. Without instrumentation you will never know which automations earned their place and which quietly broke their case in production.