AI Value Realization Theory: executive brief

01. The problem

Organizations already have AI: copilots, agents, models and pilots. But many cannot show material impact on margin, productivity, risk or social outcomes.

Central thesis: the problem is not access to AI. It is organizational conversion.

As context, MIT NANDA (2025) reports a severe gap between GenAI investment and measurable return; BCG (2025) estimates that only 5% of firms are future-built and that 60% reap hardly any material value despite substantial investment. These are context data, not an internal diagnosis.

Wrong question	Right question
How much AI do we have?	What value does it generate?
How many users?	Which decisions changed?
How many pilots?	What percentage scaled?
How much did we invest?	How much do we reinvest in capabilities?

02. The AVRT model: a conversion chain

AI → Intelligence → Decisions → Actions → Value → Reinvestment → Capabilities → More value

AI produces intelligence potential. Value appears only when that intelligence modifies decisions, is executed as actions and converts into measurable results.

In the working paper, realized value depends on three factors: AI's technical potential (P); how close the firm is to its industry's technology frontier (Γ); and the organization's conversion efficiency (Φ), which is what AVRT models. Two firms with the same P and the same Γ get different results because of Φ, and the metrics in this brief are how it is observed.

Concept	Acronym	What it measures
AI Leadership Debt Index	ALDI	Index (0–100) of the leadership and governance conditions that limit the conversion of AI into value; it works as a leading indicator
Decision Conversion Rate	DCR	Alerts that received a recorded decision, divided by the alerts that exceeded the relevance threshold
Action Conversion Rate	ACR	Actions actually executed, divided by the decisions to act
Value Conversion Rate	VCR	Actions with a positive effect relative to the baseline, divided by the actions executed
AI Reinvestment Rate	AIRR	% of captured value reinvested in capabilities

03. The weakest-link logic

The levers are complementary, not substitutes. Four levers: Capacity (talent, data, platform), Governance (accountability, risk, compliance), Leadership (alignment and incentives) and Task redesign (new roles and processes).

A weak lever cannot be compensated with excess in the others. An organization with high technology but low governance or leadership will keep capturing little value.

04. The J-curve: why the first year can mislead

At the beginning, the organization invests in training, redesign, integration and governance, but capabilities have not matured yet. That is why net value can fall before it rises.

Adjustment costs: accelerated training, consulting, process redesign, temporary productivity loss, organizational resistance, integration with legacy systems and executive focus.

Judging AI by its first year can confuse the adoption valley with failure. The question is whether the organization reinvests enough to get out of the valley.

Stage	What happens	Executive decision
0–6 months	Adjustment and learning costs	Do not overreact
6–18 months	Adoption valley	Measure ALDI, DCR, ACR, VCR
18–24 months	Exit from the valley	Sustain reinvestment (AIRR)
24–42 months	Capabilities mature	Scale and reinvest

05. ALDI: the leading indicator

ALDI is an index (0–100) of the leadership and governance conditions that limit the conversion of AI into value; it works as a leading indicator.

Financial analogy: ALDI is the stock of debt. The interest is the future loss of value and the friction to scale. AIRR is the payment or amortization of that debt through reinvestment in capabilities.

ALDI	State	Symptom	Action
0–20	Mature conversion	AI integrated into decisions	Maintain and optimize
21–40	Functional conversion	Specific bottlenecks	Resolve bottlenecks
41–60	Partial conversion	Pilots without scale	Reinvest aggressively
61–80	High debt	The frontier moves faster	Rebuild foundations
81–100	Critical debt	Shadow AI / value destroyed	Pause and redesign

06. Quick self-diagnosis: 10 questions to estimate ALDI

Score each question from 0 to 5. Add the total out of 50. Quick ALDI = 100 × (1 − Total/50). This is a preliminary executive estimate, not a formal measurement of the model.

• Capacity: dedicated AI/ML team; integrated data; MLOps/LLMOps platform.
• Governance: owner and traceability; AI committee; NIST/ISO policies.
• Leadership: common language in the C-suite; aligned KPIs; visible sponsorship.
• Task redesign: new roles for validation, orchestration or AI audit.

07. Positioning map

X axis: creation of new tasks (roles, processes, validation, audit, human-agent orchestration). Y axis: existing automation (degree of AI deployment over current tasks).

Quadrant	What it means	Recommendation
Strategic transformation	High redesign, high conversion	Maintain and scale
Local efficiency	Good redesign, partial conversion	Increase reinvestment
Automation without value	Lots of AI, low conversion	Redesign tasks and decisions
Isolated pilots	Low on everything	Invest in capacity and leadership

08. Industry cases: the critical metric changes by sector

Industry	AVRT metric	Observable evidence
Banking	DCR	Credit, fraud and pricing decisions that change because of AI; RACI, overrides
Mining	ACR	Recommendations executed as changes in operation, maintenance or shifts; OT/IT integration
Healthcare and geriatric oncology	Clinical VCR	Toxicity, hospitalization, quality of life; modified protocols, medical traceability
Retail and consumer	AIRR	Pricing/demand value reinvested in data, supply chain and capabilities
Public sector	Social VCR	Response times, access, equity, quality and accountability

09. Action roadmap: 3, 6, 12 and 18–24 months

Horizon	What to do	Deliverable
3 months	Diagnose ALDI, DCR, ACR, VCR and AIRR; identify the weakest link; set the value baseline	Position report
6 months	Redesign the weakest link, set a minimum AIRR and implement monthly monitoring	Redesigned process or decision
12 months	Scale with governance, begin the exit from the adoption valley, reduce ALDI by at least 15 points and document internal cases	First internal case with measurable value
18–24 months	Sustain AIRR and scale what works until the valley is crossed	Positive net value

10. Glossary

• ALDI — AI Leadership Debt Index: index (0–100) of the leadership and governance conditions that limit the conversion of AI into value; it works as a leading indicator.
• DCR — alerts that received a recorded decision, divided by the alerts that exceeded the relevance threshold.
• ACR — actions actually executed, divided by the decisions to act.
• VCR — actions with a positive effect relative to the baseline, divided by the actions executed.
• AIRR — percentage of generated value reinvested in future capabilities.
• Φ — the organization's conversion efficiency.
• Γ — how close the firm is to its industry's technology frontier.
• J-curve — initial valley caused by adjustment costs before value capture.

Base sources

MIT NANDA, The GenAI Divide: State of AI in Business 2025. BCG, The Widening AI Value Gap: Build for the Future 2025, 2025. BCG, AI Adoption in 2024: 74% of Companies Struggle to Achieve and Scale Value, 2024.