How to Build a Business Case for AI (NZ Guide)

A strong AI business case ties a specific use case to a measurable business outcome, estimates the full cost honestly, and sets realistic expectations for return and timeframe. For New Zealand leaders, it should also account for data-sovereignty and Privacy Act requirements, which affect both cost and approach. The best business cases are narrow and evidence-based, not sweeping "AI will transform everything" pitches.

What makes a credible AI business case?

Specificity. A credible case names the problem, the measurable outcome (time saved, cost reduced, service improved), and how you'll track it. Vague promises of transformation don't survive scrutiny from a CFO. The strongest cases start small, prove value on one use case, and use that evidence to justify the next.

What should the business case include?

• The specific problem and the measurable outcome you expect.
• Full costs — setup, licences or infrastructure, adoption and ongoing running.
• Data and governance requirements, including sovereignty and Privacy Act factors.
• A realistic timeframe and how you’ll measure success.
• Risks and how you’ll manage them, including the risk of doing nothing.

How do you handle uncertain returns?

Be honest about the range rather than inventing a precise number. Frame the first investment as a scoped, measurable pilot that proves or disproves the value at limited cost — that's far more persuasive than an over-confident forecast. A readiness assessment helps you identify which use cases have the strongest, most defensible business case.

How do you account for data sovereignty in the business case?

For a New Zealand organisation, data-sovereignty and Privacy Act requirements are not a footnote to the business case — they shape both the cost and the options. A use case that touches personal, health or regulated data may rule out the cheapest offshore tool and point toward an onshore or private deployment, which changes the cost profile from a small per-seat subscription to a larger upfront investment with predictable running costs. Surfacing that early keeps the business case honest: it is better to show leadership the real, compliant option than to win approval on a figure that quietly assumes an approach your obligations will not allow.

How do you present the cost of doing nothing?

A strong business case names the counterfactual, because standing still is not free. If staff are already using unsanctioned AI, the status quo carries real data-exposure and compliance risk. If competitors are reclaiming hours on the same knowledge work, the cost of inaction is slower service and higher unit costs that compound over time. Framing the decision as "a governed investment versus an ungoverned drift" is often more persuasive to a New Zealand board than the projected return alone, because it reframes AI from a discretionary spend into a risk that is already accruing.

Who should own the AI business case?

The strongest business cases have a business owner, not just a technology sponsor. When IT owns the case alone, it tends to be framed around the tool; when the leader of the affected function owns it, it stays anchored to the outcome — the hours saved, the service improved, the cost avoided in their own area. The ideal pairing is a business owner who feels the problem and can vouch for the value, supported by technical input on cost and feasibility and a governance voice on data and privacy. That combination is also what carries the case through a New Zealand board or CFO review, because the person presenting it can speak to the return in operational terms rather than technical ones.