COMPEL Specialization — AITE-VDT: AI Value & Analytics Expert Template 2 of 5
This template combines the business-case structure from Article 6 with the rNPV computation from Article 7. The deliverable has two components: a one-page narrative cover and a spreadsheet model. The narrative lives in this template; the spreadsheet is assembled using the structure described below.
AI Business Case: [Feature Name]
Feature: [Short name] Sponsor: [Role, name] Author: [AI value lead] Date: [Date] Version: [Draft / Final]
1. Hypothesis
Problem statement: [The business problem the feature addresses. One to two paragraphs.]
Proposed solution: [How the feature solves the problem, in feature-design terms (not technical detail). One paragraph.]
Primary business hypothesis: [Falsifiable statement of expected effect: population, metric, direction, magnitude.]
Theory of change: [The causal chain from feature operation to business outcome.]
Scope: [Boundaries of the claim — what’s in, what’s out.]
2. Investment
Build (Year 0)
| Item | Amount | Probability of completion |
|---|---|---|
| Engineering | [$] | [0.9] |
| Data preparation | [$] | [0.9] |
| Model development | [$] | [0.85] |
| Infrastructure | [$] | [0.95] |
| Total build | [$] | [0.85 compound] |
Pilot (Year 1 H1)
| Item | Amount | Probability given build |
|---|---|---|
| Pilot deployment | [$] | [0.85] |
| Measurement instrumentation | [$] | [0.9] |
| Change management | [$] | [0.9] |
| Total pilot | [$] | [0.85] |
Rollout (Year 1 H2 and beyond)
| Item | Amount | Probability given pilot |
|---|---|---|
| Broader deployment | [$] | [0.9] |
| Governance operationalization | [$] | [0.95] |
| Total rollout (Y1 H2) | [$] | [0.9] |
Annual run costs (Year 2 onward)
| Item | Year 2 | Year 3 | Year 4 | Year 5 |
|---|---|---|---|---|
| Inference cost | [$] | [$] | [$] | [$] |
| Platform | [$] | [$] | [$] | [$] |
| Governance | [$] | [$] | [$] | [$] |
| Model refresh | [$] | [$] | [$] | [$] |
| Total run | [$] | [$] | [$] | [$] |
3. Benefit
Benefit drivers
| Driver | Year 1 H2 | Year 2 | Year 3 | Year 4 | Year 5 |
|---|---|---|---|---|---|
| [Cost reduction] | [$] | [$] | [$] | [$] | [$] |
| [Revenue increase] | [$] | [$] | [$] | [$] | [$] |
| [Productivity lift] | [$] | [$] | [$] | [$] | [$] |
| Total expected benefit | [$] | [$] | [$] | [$] | [$] |
Assumption register
| Assumption | Basis | Sensitivity risk |
|---|---|---|
| [Adoption reaches X% within 6 months] | [Source or analogue] | [High / Medium / Low] |
| [Time saved per user is Y min/week] | [Pilot finding / benchmark] | [H/M/L] |
| [Value of saved hour is $Z] | [Fully-loaded wage, with caveats] | [H/M/L] |
| […] | […] | […] |
4. Risk profile
Strategic risks
| Risk | Probability | Impact | Mitigation |
|---|---|---|---|
| [Adoption shortfall] | [M] | [H] | [Change-management plan] |
| [Model capability short of threshold] | [L] | [H] | [Capability-evaluation harness] |
| [Competitive substitute emerges] | [M] | [M] | [Substitution-analysis quarterly] |
Regulatory risks
| Risk | Probability | Impact | Mitigation |
|---|---|---|---|
| [EU AI Act classification uncertainty] | [M] | [M] | [Early legal review] |
| [Data-residency constraint] | [L] | [H] | [Architecture choice] |
Operational risks
| Risk | Probability | Impact | Mitigation |
|---|---|---|---|
| [Token cost exceeds budget] | [M] | [M] | [FinOps Phase 2 practices] |
| [Drift erodes value in year 2+] | [M] | [M] | [Drift-monitoring instrumentation] |
5. Financial summary (rNPV)
Cash-flow table
| Year | Stage | Cost | Benefit | Net cash | Prob. of reaching | Expected cash | Discount factor | Discounted |
|---|---|---|---|---|---|---|---|---|
| 0 | Build | [$] | [$0] | [–$] | 1.0 | [–$] | 1.000 | [–$] |
| 1 H1 | Pilot | [$] | [$0] | [–$] | [0.85] | [–$] | 0.946 | [–$] |
| 1 H2 | Rollout | [$] | [$] | [$] | [0.77] | [$] | 0.894 | [$] |
| 2 | Operate | [$] | [$] | [$] | [0.69] | [$] | 0.797 | [$] |
| 3 | Operate | [$] | [$] | [$] | [0.66] | [$] | 0.712 | [$] |
| 4 | Operate | [$] | [$] | [$] | [0.64] | [$] | 0.636 | [$] |
| 5 | Operate | [$] | [$] | [$] | [0.62] | [$] | 0.567 | [$] |
| rNPV | [$] |
rNPV sensitivity
- Base case: [$]
- p10 (Monte Carlo, 10,000 iterations): [$]
- p50: [$]
- p90: [$]
- Probability of positive rNPV: [%]
- Probability above hurdle ($X): [%]
Tornado chart — three most-sensitive inputs
- [Input 1]: impact range [$X] to [$Y]
- [Input 2]: impact range [$X] to [$Y]
- [Input 3]: impact range [$X] to [$Y]
Payback
- Cumulative-benefit-equals-cumulative-cost date: [Year, quarter]
- Payback period: [Years]
6. Recommendation
Recommendation: [Proceed / proceed with conditions / do-not-proceed]
Conditions (if applicable):
- [Condition 1]
- [Condition 2]
- [Condition 3]
Rationale: [Two to three sentences connecting the rNPV, the sensitivity, and the risk profile to the recommendation.]
Stage-gate requested: [Calibrate exit / Organize exit / etc.]
Decision date requested: [Date]
Decision makers: [Named roles]
Appendix A — Computational notes
Discount rate: [X%] Justification: [Cost of capital, risk adjustment, basis.]
Probability chain: Build probability × pilot-given-build × rollout-given-pilot × operate-given-rollout. Probability does not multiply across operating years — each operating year uses the cumulative probability of reaching rollout.
Monte Carlo distributions: [List distributions used for each varying input with parameter values.]
Assumption sensitivity: [List which assumptions were varied, what ranges, and how the rNPV responded.]
Appendix B — Linkage to other artifacts
- Measurement plan: [Reference]
- KPI tree: [Reference]
- Compute budget: [Reference]
- Risk register: [Reference]
Appendix C — Sign-offs
| Role | Name | Date |
|---|---|---|
| AI value lead | [Name] | |
| FinOps lead | [Name] | |
| CFO / sponsor | [Name] | |
| Program office | [Name] |