Value realization
Whether the money produced what the business case said it would.
Value realization connects a business case to measured outcomes — establishing what an investment was supposed to produce, tracking whether it did, and maintaining accountability for the gap. It is the part of the investment cycle most organizations skip, because approval depends on the business case and nothing depends on verifying it.
Most disputes about whether an investment paid off are not disagreements about the investment. They are the predictable result of deploying without a baseline, which makes the question unanswerable rather than merely contested. By the time anyone asks, the programme has disbanded and the counterfactual is gone.
Who this is for
Primarily the CFO, transformation leadership, PMO and portfolio leadership, and the CIO — the people who have to defend a technology run-rate to a board, an investment committee or an oversight body, and who currently cannot evidence what it bought.
The chain that has to hold
Value realization fails at whichever link is missing, and in most organizations more than one is:
Why the cadence matters
Benefits are typically assessed at gate reviews and then at a post-implementation review that frequently never happens. The problem is that assumptions decay faster than benefits arrive: a business case rests on a dozen assumptions about volume, adoption, headcount and timing, and several will be wrong within two quarters.
Where the chain from assumption to expected benefit is kept live, a missed benefit is diagnosable — you can see which assumption moved. Where it exists only in an approved paper, the organization is left arguing about attribution, and usually stops asking.
The AI-specific requirement
For conventional technology, usage is a reasonable proxy for effect. For AI it is not: a system can produce large volumes of output nobody acts on while adoption metrics look healthy. What is needed instead is a record of which outputs entered a decision — which recommendations were approved, by whom, and what followed. That is a provenance requirement, and it is why Tacit OS attaches a chain of custody to every output.
How EXOS runs it
What you keep
The live chain from business case through assumption to measured outcome. When a benefit misses, the question is which assumption broke — answerable only because they were recorded as something other than prose in an approved paper.
Evidence
A federally funded research programme translated scientific impact into the fiscal accountability language federal evaluators required, and survived fiscal review. Healthlink Advisors made methodology measurable enough to serve as an acquisition asset. Both are in the case library.
Delivered through fdX
EXOS delivers this through fdX — Forward-Deployed Expertise. Here the expertise is finance, portfolio and programme judgment — people who have defended an investment to a board. Engineering instruments the measurement chain so the baseline is real, and Tacit OS keeps the link from assumption to outcome live rather than frozen in an approved paper.
Common questions
How do you measure value from enterprise AI?
The same way as any other investment, with one addition: a baseline set before deployment, a named owner for each claimed benefit, a cadence that outlasts the project, and — specific to AI — provenance showing which outputs were actually acted on. See how to measure value from enterprise AI.
Can this start after a programme has finished?
Partially. Baselines can sometimes be reconstructed from historical data, but reconstruction is weaker evidence than measurement and will not satisfy a sceptical board. Starting late costs rigour — it is still worth doing, and worth not repeating.