01 / Section
Start by confirming the process repeats
Repeated work creates a reason to look closer. It does not establish that the process is stable, safe, valuable, or technically suitable for automation.
The first useful threshold is evidence of at least two distinct workflow attempts with aligned goals and actions. From there, a person can inspect the differences and decide whether the pattern is real enough to assess.
02 / Section
Separate evidence from estimates
Observed steps can support a description of the current process. They do not automatically establish hours saved, implementation effort, or financial return. Those conclusions require explicit assumptions about workload, cost, risk, and the part of the process that could change.
Keeping recorded evidence and planning assumptions distinct makes the opportunity easier to challenge and update.
03 / Section
Make assumptions visible
Financial estimates should state the personal workload and cost inputs that produced them. Draft assumptions are not confirmed facts, and an estimate can become stale when volume or the process changes.
Labels such as draft, confirmed, and stale help reviewers understand what they can rely on without flattening every number into the same level of confidence.
04 / Section
People decide what gets automated
Shadow can surface a repeated process and suggest that it deserves review. It does not execute the change. A person evaluates the evidence, the exceptions, the risks, and the desired outcome.
This keeps automation review separate from execution.