Value Realization
Projected vs. Realized AI ROI: Why They Diverge
A framework for comparing the value forecast in an AI business case with the value actually realized after deployment.
Projected ROI is an investment hypothesis
Before implementation, an AI business case estimates what could happen if a set of assumptions is true.
Realized value depends on actual usage
Track active users, eligible transactions, actual utilization and the percentage of work that genuinely moves through the new process.
Time saved is not automatically financial value
Hard savings require a real change in spending. Capacity value requires evidence that released time is used productively.
Production costs change the denominator
Realized ROI should use the actual ongoing cost of operating the capability.
Benefit leakage explains many gaps
Measure the complete workflow rather than only the AI interaction.
Build a forecast-to-actual bridge
Show the projected assumption, actual result and financial effect of each major variance.
Use realized value to decide the next investment
Realized-value measurement turns AI governance into a continuing capital-allocation process.
Common questions
Why do projected and realized AI ROI diverge?
The forecast assumes a level of adoption, a benefit per use, and a cost to operate. After launch, usage, workflow leakage, and production cost are often different. Realized ROI uses those actuals. The original figure stays useful as the hypothesis you are testing.
What should you track to explain the gap?
Track active users, eligible transactions, actual utilization, and the share of work that moves through the new process. Record the ongoing cost, including integration, support, monitoring, and model usage. Then put the projected assumption, the actual result, and the financial effect of each variance in one bridge.
When does saved time count as realized value?
It counts when you can show what happened to the time. Hard savings need a real change in spending. Capacity value needs evidence that the released time is used for more work, faster work, or work that would otherwise have cost more. A time saved figure on its own is not a financial result.
How should this comparison be used after approval?
Run it as a routine, not a one-time slide. The AI value realization framework is that routine: baseline, owner, adoption, total cost, and a decision to scale, adjust, or stop. Enterprise AI ROI applies the same comparison once a pilot moves across functions.
What decision should the gap change?
Use it to choose the next investment. If adoption is the constraint, fix the workflow before spending more on the model. If production cost moved the return, update the AI ROI model before you scale. If the pilot never had a baseline, start with the measurement steps in Why Most AI Pilots Never Show Measurable ROI.