Choose a representative baseline
Record enough current work to capture normal volume and exceptions. Measure how long routine preparation takes, how long requests wait and where records are incomplete. Use the same definitions before and after implementation.
Separate speed from business value
A faster response can be useful without proving higher revenue. Track the operational measure first, then examine whether it influences the business outcome. Keep attribution and other process changes visible so the system does not get credit for every improvement.
Include the full operating cost
Include implementation, platform usage, ongoing management, employee reviews and recovery work. For a simple planning estimate, multiply task volume by average handling time to understand current effort. Treat that estimate as an assumption until measured.
Agree on output quality
A completed run can still produce an unusable record. Review completeness, source accuracy, duplicates, handoff ownership and whether the employee can act. Include exception and failure rates alongside completion and response time.
Report the limitations
Distinguish observed changes, estimates and unresolved issues. State the period measured and the conditions that changed. A useful report tells the client what improved, what remains manual and what needs attention next.
