Why this dimension matters
Value is the only Trust & Performance dimension that the CFO reads. Without a credible Value number, every other dimension becomes a cost center to be cut in the next budget cycle. Conversely, a credible Value number funds the reliability, safety, and governance work that makes the system trustworthy in the first place. Value measurement is the oxygen supply of the transformation program.
Core metrics
Metric 1: ROI (return on investment)
Definition. Net benefit (value captured minus total cost of ownership) divided by total cost of ownership, for a named use case across a defined measurement window.
Formula. roi = (value_captured − tco) / tco.
Cadence. Measured at 90, 180, and 365 days post-go-live, then annually.
Owner. Business sponsor with finance validation.
Deep dive. ROI quantification is covered in depth in M2.5Business Value and ROI Quantification, including counterfactual baselines, attribution discounting, and the handling of soft benefits.
Metric 2: Business-outcome attainment rate
Definition. The percentage of committed business outcomes (documented in the initiative’s value case) that have been achieved at the measurement point.
Formula. attainment = (outcomes_achieved / outcomes_committed) × 100, weighted by outcome dollar value.
Cadence. Quarterly.
Owner. Portfolio owner.
Why separate from ROI. ROI measures dollars; attainment measures promises kept. A program with high ROI but low attainment is accidentally successful — and accidents do not compound. A program with high attainment is a program whose forecasts can be trusted, which is what compounds into executive confidence.
Metric 3: Productivity uplift
Definition. The percentage change in throughput, cycle time, or error rate for a defined work unit, measured against a pre-AI baseline.
Formula. uplift = ((post_metric − pre_metric) / pre_metric) × 100.
Cadence. Baseline at go-live; re-measured quarterly.
Owner. Process owner with HR/operations.
Honesty rule. Productivity uplift must be measured on the same work population pre- and post-deployment. Comparing a new team on a new process to an old team on an old process is not a productivity measurement — it is a morale exercise.
How to measure
- Write the value case before the build. Commit to a dollar number, a time horizon, and a method for measuring it.
- Capture the pre-AI baseline on the same metric the post-AI system will produce. A baseline written after the fact is a rationalization, not a measurement.
- Run a counterfactual or holdout wherever possible. Where impossible, document the attribution assumption explicitly.
- Report all three metrics together on the trust scorecard. Any one in isolation can be gamed.
- Reconcile annually with finance. If finance cannot see the dollars on the P&L, the ROI is a slide, not a number.
Targets and thresholds
- ROI. Program-dependent; publish the target in the value case. Break-even by month 12 is a common minimum for enterprise initiatives.
- Outcome attainment. Above 70% is healthy; above 85% is mature.
- Productivity uplift. Use-case dependent; publish the target in the value case.
Common pitfalls
Counting the same dollar twice. Productivity uplift that is already captured in ROI should not be reported as additional value.
No baseline. Without a pre-AI baseline, there is nothing to subtract from the post-AI number.
Attribution claiming. A sales lift in the same quarter as an AI deployment and a new pricing strategy belongs partly to both — report the assumption.
Soft benefits as hard numbers. Employee satisfaction, brand uplift, and reduced risk are valid but belong in a separate column from cash ROI.
Related articles
Module 2.5, Article 04: Business Value and ROI Quantification M2.5Value Realization Reporting and Communication M4.1Portfolio Value Realization and Benefits Tracking M3.6The Measurement and Value Realization Framework M2.5Measuring AI Adoption