Policy : Prioritise AI Use Cases by Impact, Not Novelty

Commitment to Impact-Driven AI Prioritisation The single greatest source of wasted AI investment is the selection of use cases based on what is technically interesting rather than what is organisationally valuable. Teams build AI systems for problems that were already adequately solved, for use cases where the AI adds marginal value over a simpler solution, or for applications that generate compelling demonstrations but do not survive contact with real operational conditions. Our commitment is to apply rigorous prioritisation discipline to AI use cases — selecting work based on expected impact, user need, and strategic importance, not on the appeal of the technology or the novelty of the application.

What This Means Impact-driven prioritisation means building an AI use case pipeline with defined evaluation criteria and systematically assessing candidates against those criteria before committing resources. It means asking the hard questions: Is this problem worth solving with AI? Is AI the right tool for it? What is the realistic business impact if it works? What is the effort and risk involved? And it means being willing to discard use cases that do not meet the bar — including ones that are technically exciting, publicly visible, or championed by senior stakeholders.

Our commitment to prioritising AI use cases by impact over novelty is built on:

• Structured Use Case Evaluation – AI use case candidates are evaluated against a defined set of criteria before entering the delivery pipeline. Criteria include business impact potential, user need strength, data availability, technical feasibility, and risk profile. Informal selection driven by enthusiasm or seniority is replaced by structured assessment.
• Impact Estimation Requirements – Every use case candidate must include an impact estimate: the expected magnitude of the business outcome improvement, the number of users or transactions affected, and the confidence level of the estimate. Unquantified impact claims are not sufficient to justify project initiation.
• Comparative Prioritisation – Use cases are prioritised relative to each other, not just assessed in isolation. This forces explicit trade-offs between competing opportunities and prevents the AI portfolio from filling with medium-impact work simply because it is easy to initiate.
• Simpler Alternative Assessment – Before committing to an AI solution, teams assess whether a simpler, non-AI approach could meet the user need at lower cost and risk. AI is selected when it provides meaningful advantage over simpler solutions — not by default because it is available.
• Strategic Alignment Screening – Use cases are screened for alignment with organisational strategy. AI investment is directed toward the areas of greatest strategic importance, not distributed across every department that expresses interest.
• User Need Validation – Use case selection is informed by validated user need, not assumed demand. Teams talk to the people who will use or be affected by the AI system before committing to building it. Discovery precedes delivery.
• Regular Portfolio Rebalancing – The AI use case portfolio is reviewed regularly to ensure it remains focused on the highest-impact opportunities. As business context changes, use case priorities are updated — AI project queues are not first-in-first-out backlogs frozen at initiation.

Why This Matters The most common failure mode in enterprise AI is not technical — it is strategic. Organisations build technically impressive systems that solve the wrong problems, duplicate capabilities that already exist, or address use cases with insufficient scale to justify the investment. Novelty is a seductive but unreliable guide to value. Impact-driven prioritisation is the discipline that ensures the organisation's AI investment is concentrated where it can genuinely matter.

Our Expectation Every AI initiative entering delivery has passed a structured impact assessment and has been prioritised relative to competing opportunities. Teams that self-initiate AI projects based on technical interest without going through prioritisation are not being entrepreneurial — they are making unilateral resource allocation decisions. Prioritising AI by impact, not novelty, is how we ensure the organisation's AI capability delivers genuine, concentrated Value.