Beyond AI ROI: Why Value Realization Depends on the Operating Model

The enterprise conversation around artificial intelligence is changing.

For several years, much of the discussion centered on capability: What can AI do? Which use cases should we pursue? How quickly can we move from pilot to production?

The harder question now is:

Where is the value?

Not the projected value or the productivity claim embedded in the business case. Where is value actually showing up in the enterprise?

Implementation and value realization are not the same thing. An AI capability can be deployed successfully, employees can use it, and a program can reach its milestones while the enterprise still fails to realize the value that justified the investment.

This is not exclusively an AI problem. It is a strategic execution problem that AI is making easier to see.

The AI value conversation is becoming an operating-model conversation

Recent 2026 research reflects the shift.

McKinsey has argued^[1] that AI value is increasingly tied to operating-model choices: how work moves across functions, how decisions are made, how teams are organized, and how the organization changes the work around the technology.

BCG has similarly emphasized end-to-end redesign^[2] rather than layering AI onto existing work, while Gartner has framed AI ROI as a question of value rather than feasibility alone^[3]. Oliver Wyman has also described agentic AI scale as an operating-model issue^[4] involving governance, funding, talent, delivery, and economics.

The evidence types differ, and none validates any single transformation framework. But together they reinforce a practical executive concern:

Getting AI into production and converting AI into enterprise value are different problems.

So instead of asking only, What is the ROI of this AI investment?, leadership should also ask:

What must change in the business for the claimed value to become possible?

ROI sits at the end of a much longer realization chain

ROI should be assessed before investment and revisited throughout delivery. Organizations should expect capital to produce an appropriate return.

But realized ROI is an outcome of value realization. It is not a substitute for understanding how value will be realized.

A business case may state that an initiative will reduce cost, create capacity, improve conversion, reduce risk exposure, or increase margin. Those claims are useful, but they are not yet operating reality.

Something inside the enterprise must move first: throughput, exception volume, decision latency, customer effort, control reliability, authority, data availability, functional capacity, or a cross-functional handoff.

In most consequential initiatives, those changes do not occur inside one team. They move across a value stream.

That is why value-stream thinking matters to value realization.

A technology investment should not be examined only through the system being implemented. It should be examined through the end-to-end flow of work, decisions, information, controls, capabilities, and functional contributions through which the promised result must eventually appear.

The outcome owner belongs at the beginning, not the end

The functional executives accountable for the result must help shape the initiative from the outset. They bring current operating evidence, define the contribution their functions must deliver, and surface constraints, acceptance conditions, decision rights, and capability requirements before assumptions harden into implementation commitments.

In S.C.A.L.E., these leaders are the business outcome owners: executives accountable for the measurable contribution their functions must deliver and sustain so an enterprise remit can become operating reality.

Their responsibility begins before go-live and continues after implementation support leaves.

The Value-Realization Chain

I use the following practitioner-developed chain to trace and test a value claim:

The chain is a reasoning structure. It is not proof of causality, nor is it a mandatory linear project lifecycle. Evidence must support the connection between each step.

For financial claims, apply Finance-agreed recognition rules. For customer, risk, compliance, resilience, or other outcomes, use the agreed measures and accountable authorities appropriate to those claims. Be explicit about what has been monetized and what has not.

Every link matters. If the strategic claim never translates into the affected value stream, the initiative remains abstract. If required functional contributions are unclear, accountability remains vague. If the operating model cannot support the change, implementation creates workarounds. If behaviors, decisions, controls, and measures remain unchanged, the old operating model continues to govern the business.

And if improvement disappears after implementation support is removed, the enterprise experienced a temporary performance event, not durable value realization.

Eight questions I would ask before declaring an AI initiative positioned to realize value

In 100 Questions Every Business Outcome Owner Must Ask, the questions follow the life of an enterprise outcome, from strategic relevance and current operating truth through readiness, governance, realization, and sustainment.

They are not an AI implementation checklist. For an organization examining whether an AI initiative is positioned to create value, however, I would begin with these eight.

Q1. What disruptive business trend or competitive change is prompting this initiative?

This protects against starting with the technology.

The organization should be able to identify the enterprise condition requiring change, the limitation in the current operating environment, where that limitation exists in the value stream, and what value is exposed if nothing changes.

What it tests: Whether a defined business condition warrants action.

Q4. What measurable change would meaningfully address the business challenge or growth opportunity?

One of the easiest transformation mistakes is measuring new activity without establishing a new operating truth.

More AI interactions, greater tool adoption, additional agents deployed, more prompts submitted, or faster task completion may all be useful indicators.

None automatically establishes enterprise value.

The better question is: What will the organization be able to observe, decide, or accomplish differently if this initiative works?

What it tests: Whether the required improvement is measurable and relevant.

Q6. Which value streams, spanning multiple functions, must convert this enterprise remit into business reality?

Enterprise value rarely respects the organization chart.

A customer outcome may cross Sales, Operations, Technology, Finance, and Customer Service. A productivity or risk claim may depend on changes across data, policy, controls, workflows, skills, systems, and authority.

Mapping participating teams is not enough. Leadership must understand how the enterprise remit moves through connected value streams, where value can fracture, and who protects performance at those boundaries.

What it tests: How the enterprise remit depends on connected value streams.

Q7. What business capabilities must be evaluated within each value stream to establish the conditions required to realize return on investment?

Once the value streams are understood, the next question is whether the enterprise possesses the capabilities required to operate differently.

AI does not eliminate capability requirements; it changes them. Data quality, decision rights, process integration, architecture, skills, governance, controls, measurement, vendor capability, and operating discipline may determine whether the conditions required for return exist.

What it tests: Which capabilities must support the intended return.

Q14. Who currently owns value delivery - and where does binding decision authority actually sit?

Projects usually have leaders. Value does not always have an owner.

Sponsorship, program leadership, technology ownership, functional accountability, and binding decision authority are not interchangeable. When trade-offs arise across functions, leadership must know who owns value delivery and who can actually authorize the decision required to protect the outcome.

What it tests: Where current value ownership and binding decision authority sit.

Q36. Have we defined success in quantifiable enterprise-value terms rather than delivery milestones?

A completed implementation establishes that work was delivered. It does not establish that the enterprise condition leadership intended to change actually changed.

The value claim needs an evidence chain connecting operating movement, performance outcome, the value mechanism, applicable validation or recognition rules, and the sustainment condition.

For financial claims, Finance must be able to validate the economics. Other value classes require their own governed measures and authorities.

What it tests: Whether success is defined through enterprise value.

Q38. How will adoption, capability, performance and impact be measured separately?

These concepts are routinely collapsed.

They should not be.

Usage is not adoption. Adoption is not capability. Capability is not performance. Performance is not automatically enterprise value.

Opening an AI tool proves access or use. It does not establish correct capability, workflow change, improved performance, downstream benefit, or enterprise value.

What it tests: Whether adoption, capability, performance, and impact have distinct evidence.

Q77. What operational dashboards or reporting views will confirm sustained value after go-live?

Value realization should survive the implementation organization. If the evidence disappears when the transformation team leaves, the operating model has not fully inherited the outcome.

Sustained value requires ordinary management systems capable of revealing operating performance, cross-functional outcomes, value, capability health, and deterioration over time.

What it tests: Whether normal operating reporting can reveal sustained value.

The business case and the value-realization case answer different questions

These are complementary lines of inquiry. In a mature organization they may live within the same governance artifact rather than two separate documents.

What matters is that both questions are answered.

This distinction matters as organizations accelerate AI investment. Technology can move faster than the business can redesign itself around it, creating an illusion of progress through more deployments, agents, automated tasks, and adoption dashboards.

Enterprise transformation is not measured by the volume of new capability introduced. It is measured by whether the enterprise converts that capability into an outcome it can prove, validate, govern, and sustain.

AI is today's pressure test. The discipline applies to every strategic initiative.

The same reasoning applies beyond AI.

ERP transformations. Operating-model redesign. Digital modernization. Cost transformation. Customer-experience initiatives. Mergers and integrations. Shared-services transformations. Regulatory programs. Supply-chain transformations. Growth initiatives.

Any strategic initiative that makes a claim about future enterprise value creates the same obligation:

Trace the claim into the operating reality required to produce it.

That means understanding where value must move, which functions must contribute, what capabilities must exist, which decisions must change, what evidence proves movement, who owns the outcome, and what must endure after implementation.

That is the work of value realization.

A practical place to begin

100 Questions Every Business Outcome Owner Must Ask was created for these moments.

It is not intended to be read as 100 questions in sequence or as another transformation checklist. It is a working executive guide for entering an initiative where greater clarity is required, whether the issue is strategy, baseline integrity, value streams, readiness, risk, adoption, governance, ROI, or sustainment.

Each question expands into a working chapter with a model, evidence guidance, decision standard, accountability, and an output blueprint.

The method is straightforward:

Start with the decision that needs to be made.

Ask the question capable of exposing the operating truth behind it.

Source the evidence.

Clarify ownership and authority.

Then determine whether leadership has a position it can responsibly act on.

Because the objective is not to produce better answers for a presentation.

It is to make the conditions required for value visible before the value is placed at risk.

Inside the Guidebook

From question to executive output

Find the question or outcome relevant to your initiative. Use its chapter and blueprint to examine what matters, clarify what is required, and develop an output leadership can act on.

Each question expands into a working chapter:

1. 01The primary executive question
2. 02A model or framework for examining it
3. 03Guidance on where to source the evidence
4. 04An executive decision standard
5. 05The accountability and authority lens
6. 06An Executive Output Blueprint

Explore the output index↗Executive Outputs Index · 2 of 10 · Guidebook page 500

Read the blueprint↗Question 16 · Future-State Cross-Functional Operating Contract · Guidebook page 108

Explore 100 Questions Every Business Outcome Owner Must Ask by Torie T. Thomas, or get the complimentary First Digital Edition through the existing guidebook delivery experience.