Keep ROI models understandable. A simple model with visible assumptions is more useful than a complex model nobody trusts.
Labour capacity
Annual hours released = annual volume × minutes saved per item ÷ 60
Capacity value = annual hours released × loaded hourly cost
Then ask how the capacity will actually be used. If no headcount is removed, frame this as capacity rather than guaranteed cash saving.
Avoided hiring
Avoided hiring value = FTE requirement avoided × loaded annual FTE cost
Use only where the organisation genuinely expects to add capacity absent the change.
Error reduction
Annual benefit = current error volume × cost per error × expected reduction
Include downstream rework, credits, complaints or operational cost only where supported.
Revenue opportunity
Expected revenue benefit = relevant volume × conversion uplift × average contribution per conversion
Contribution is often more useful than headline revenue when costs scale with sales.
Payback
Payback period = implementation investment ÷ monthly net benefit
Simple ROI
ROI = (annual benefit − annualised cost) ÷ annualised cost
Sensitivity analysis
Change the two or three assumptions that drive the model most. Typical sensitivity variables include adoption, minutes saved, transaction volume and implementation cost.
Do not present Iris-generated or consultant-estimated ROI as realised benefit. The diagnostic establishes an expected case; realised value must be measured after implementation.
Minimum assumptions table
For each input include: source, owner, current value, scenario value, confidence and review date. 
