Policy : Align AI Investment to Measurable Outcomes

Commitment to Outcome-Aligned AI Investment AI is an exceptional technology that can genuinely transform how organisations operate and deliver value. It is also one of the most easily misallocated investments in the modern enterprise. The enthusiasm around AI capabilities creates pressure to build AI systems simply because we can, because competitors are doing it, or because the technology is genuinely fascinating. Our commitment is to resist that pressure and insist that every significant AI investment is anchored to specific, measurable outcomes that matter to the organisation — outcomes that we track, report on, and use to make decisions about continued investment.

What This Means Aligning AI investment to measurable outcomes means defining what success looks like before the first line of code is written or the first model is trained. It means identifying the business metric that will improve, by how much, for whom, and in what timeframe. It means building measurement into the delivery plan from day one, not bolting on an evaluation framework after the system is live. And it means being willing to stop AI projects that are not delivering against their stated outcomes, rather than continuing to invest in the hope that value will eventually materialise.

Our commitment to outcome-aligned AI investment is built on:

• Outcome Definition Before Commitment – Before any AI project receives significant investment, the team defines the specific business outcome it will improve. This outcome is expressed in terms that can be measured: reduction in processing time, increase in conversion rate, decrease in error rate, improvement in user satisfaction score. Vague outcome statements such as "improve efficiency" or "enhance the customer experience" are not accepted.
• Baseline Measurement – Every AI project establishes a baseline measurement of the outcome it intends to improve before building begins. Without a baseline, claims of improvement are unverifiable and investment decisions are made on faith rather than evidence.
• Success Thresholds and Review Triggers – Projects define minimum success thresholds: the level of outcome improvement that justifies the investment. They also define review triggers — points at which progress against outcome metrics is assessed and decisions about continuation are made explicitly.
• Attribution Discipline – We are rigorous about attributing outcome changes to AI systems rather than to concurrent business changes, seasonal effects, or other factors. Correlation between AI deployment and outcome improvement is not sufficient — we invest in the measurement infrastructure needed to establish causal attribution where it matters.
• Portfolio-Level Outcome Tracking – AI investment is tracked at portfolio level, not just project level. Leadership has visibility of what the aggregate AI investment is delivering in terms of measurable business outcomes, enabling resource allocation decisions to be made on evidence.
• Sunset Criteria – Every AI project has defined criteria under which it will be scaled back, redesigned, or shut down if it does not deliver. This prevents the indefinite continuation of AI projects that are technically interesting but not delivering organisational value.
• Value Realisation Reviews – After deployment, AI systems undergo formal value realisation reviews at defined intervals. These reviews assess actual outcome delivery against the original investment thesis and inform decisions about ongoing investment in the system.

Why This Matters Organisations that invest in AI without clear outcome alignment consistently discover that their AI portfolios are technically impressive and strategically inconsequential. The absence of measurable outcomes creates a accountability vacuum — teams cannot demonstrate success, leaders cannot justify continued investment, and AI initiatives drift from strategic tools into expensive proof-of-concept graveyards. Measurable outcomes are the mechanism by which AI investment earns ongoing organisational trust and resources.

Our Expectation Every AI project above a defined investment threshold has a documented outcome definition, a baseline measurement, and a defined measurement plan. Projects that cannot articulate what organisational outcome they are improving are not approved for significant investment. Aligning every AI investment to measurable outcomes is how we ensure our AI work delivers genuine Value.