Framing the Use Case

Be specific about scope

A good ROI case starts with a specific process (processing supplier invoices), its owner, its volume, today's cost and quality, and the change proposed (AI extracts fields and matches purchase orders; staff review). Vague scopes ("use AI in finance") cannot be estimated or measured. Define what success means operationally (minutes per invoice, error rate, cycle time) and how you will measure it before and after.

A use-case framing card

One card per proposed automation.

process:  supplier invoice processing (accounts payable)
owner:    AP team lead
volume:   ~12,000 invoices / month
today:    6.5 min per invoice, 4% error rate, 3-day cycle time
proposal: AI extracts fields + matches PO; staff review flagged items
success:  minutes per invoice, error rate, cycle time
measure:  timing sample + ERP logs, before and after, with a comparison team

Start with high-volume, rule-heavy work

Processes with large volume, clear rules and measurable outputs give the most reliable ROI estimates.

Quick check: Why is "use AI in finance" a poor scope for an ROI estimate?

• ROI only applies to marketing
• Finance cannot use AI
• It is too small
• It is too vague to measure a baseline or the change