Standard : AI use cases are selected based on validated business impact
Purpose and Strategic Importance
This standard requires that AI use cases must demonstrate validated evidence of business impact — through data analysis, user research, or a structured feasibility assessment — before engineering investment is approved. It supports the policy of aligning AI investment to measurable outcomes by ensuring that resources flow toward problems where AI can make a genuine, quantifiable difference. Use cases selected on the basis of novelty, vendor enthusiasm, or executive sponsorship alone consistently underdeliver on value.
Strategic Impact
- Concentrates scarce AI talent and infrastructure on the opportunities with the highest return on investment
- Creates a shared language between business and engineering for discussing AI priorities in terms of outcomes
- Reduces the frequency of AI projects that are technically successful but deliver no measurable business benefit
- Builds a prioritised pipeline of AI opportunities that reflects real organisational needs rather than technology fashion
- Enables portfolio-level visibility into the expected and realised value of AI investment across the organisation
Risks of Not Having This Standard
- Engineering effort is wasted on AI use cases that address low-impact problems or problems that do not exist at scale
- Business stakeholders lose confidence in AI delivery after repeated projects that produce demos but not value
- The organisation builds AI capability in the wrong areas, creating misaligned technical debt
- High-value problems remain unsolved because they were not surfaced through a structured selection process
- Budget for AI is reduced organisation-wide after the portfolio fails to demonstrate a credible return
CMMI Maturity Model
Level 1 – Initial
| Category |
Description |
| People & Culture |
- Use cases are proposed informally based on individual enthusiasm or vendor pitches with no business validation |
| Process & Governance |
- No selection framework; projects start when a sponsor secures budget, regardless of evidence of impact |
| Technology & Tools |
- No tooling to assess or compare use case value; decisions are made in ad hoc meetings |
| Measurement & Metrics |
- No expected value is defined at project inception; success criteria are vague or absent |
Level 2 – Managed
| Category |
Description |
| People & Culture |
- Teams are expected to write a brief business case before starting AI work; product and business owners are involved in scoping |
| Process & Governance |
- A simple use case intake form captures problem statement, affected volume, and estimated impact before work begins |
| Technology & Tools |
- A shared backlog or register of AI opportunities is maintained; items are ranked by estimated business value |
| Measurement & Metrics |
- Each approved use case has a stated target metric and baseline; progress against the metric is reviewed at milestones |
Level 3 – Defined
| Category |
Description |
| People & Culture |
- A cross-functional panel (AI, business, data, risk) evaluates use cases against a defined impact framework before investment is approved |
| Process & Governance |
- A structured use case evaluation framework scores feasibility, data availability, strategic alignment, and estimated value |
| Technology & Tools |
- A value tracking tool links approved use cases to their KPIs; realised value is tracked from pilot through to production |
| Measurement & Metrics |
- Use cases are evaluated against a quantified impact threshold; those below the threshold are deprioritised or reshaped |
Level 4 – Quantitatively Managed
| Category |
Description |
| People & Culture |
- Investment decisions are informed by a portfolio view of expected versus realised value; teams are accountable for delivering against stated impact estimates |
| Process & Governance |
- Use case selection is reviewed quarterly; cases that fail to demonstrate impact within defined timeframes are deprioritised |
| Technology & Tools |
- Predictive models inform use case prioritisation based on historical delivery data and feasibility signals |
| Measurement & Metrics |
- ROI per use case is calculated at project close; prediction accuracy of impact estimates is tracked to improve future assessments |
Level 5 – Optimising
| Category |
Description |
| People & Culture |
- Teams contribute retrospective impact data that continuously improves the organisation's ability to predict AI value |
| Process & Governance |
- The use case selection framework is continuously refined based on delivery outcomes and changes in business strategy |
| Technology & Tools |
- AI-assisted opportunity discovery tools surface high-potential use cases from operational data and process analysis |
| Measurement & Metrics |
- Use case selection accuracy and portfolio-level value realisation are benchmarked against industry peers |
Key Measures
- Percentage of AI projects with a documented and validated business case before engineering starts
- Average predicted versus realised business impact across completed AI use cases
- Proportion of AI projects that achieved their stated impact target within the defined timeframe
- Number of use cases deprioritised per quarter due to failure to meet impact threshold in the evaluation framework
- Portfolio-level ROI from AI investment measured annually against baseline year