Automation ROI math fails when the baseline is missing or the assumptions stay hidden. Use inputs a buyer can inspect and recompute.
The usual version multiplies your headcount by an hourly rate by a percentage somebody invented, and the result is a number nobody in your company can defend in a budget meeting. When the project ships and the savings show up nowhere, the whole category loses credibility. The fix is checkable inputs, assumptions kept visible, and the same definitions before and after the build.
Start with a baseline you measured, not estimated. Count the workflow's real volume for a real period: how many times it ran last month, how long each run takes, where it waits, and how often a person had to redo or rescue the output. The rescue rate matters as much as the run time. A task that takes six minutes but comes back wrong a fifth of the time costs more than six minutes.
Write the assumptions where the reader can see them. Scenario example: “800 messages a month at 6 minutes each is 80 hours. If 70% become 4-minute reviews, that is roughly 37 hours of modeled capacity.” Anyone can check that arithmetic, challenge the 70%, and rerun it with their own number.
Call a scenario a scenario. Before the build, every number is modeled, not measured. Say so on the page. Buyers notice when the word is missing.
Measure after with the same definitions as before. If the baseline counted rework time, the after counts rework time. Keep the definitions, period length, and units consistent.
Keep the misses in the record. Some builds do not clear the bar. A workflow that saved four hours a month against a real cost to build and run should be shut down, and the readout should say so. A measurement practice that never produces a negative result is not a measurement practice.
My measurement standard records the before and after in the same units. A workflow that misses its threshold gets changed or stopped. Every scenario on this site puts its assumptions and arithmetic on the page.
The KPI & Operations Reporting Engine exists partly because this measurement work is itself a recurring workflow most teams do by hand.
The free written assessment is the smallest version: you describe the drag, I send back the math I would run on it.