A Business Case That Survives Scrutiny

Be honest rather than optimistic, because this is the input that most affects credibility. For a well-scoped initiative using proven approaches in an organisation with relevant experience, seventy to eighty-five per cent is defensible. For a first AI project in an organisation with limited data maturity, fifty to sixty-five per cent is more realistic. Below fifty per cent, you are describing an experiment rather than an investment, and it should be framed and funded as one, a small pilot with a decision point rather than a full business case.

Assume you capture forty to sixty per cent of full-year benefit in year one for a typical initiative. That reflects a build period of one to three months, followed by several months during which adoption climbs and people are still adjusting their working patterns. Initiatives requiring behaviour change from many people ramp more slowly than those operating in the background. The most common business case error is assuming full benefit from month one, which overstates year one by roughly half and is immediately obvious to anyone reviewing it.

Mention them, but keep them out of the numbers. A case built on quantified hard benefits with soft benefits described qualitatively alongside is far more persuasive than one that assigns a dollar figure to staff satisfaction. Reviewers discount soft benefits heavily and, worse, their presence in the calculation invites scepticism about the hard numbers too. Let the financial case stand on measurable ground and let the soft benefits be the reason to prefer this initiative over an equally profitable alternative.

Be explicit about the mechanism rather than assuming the value. If saved capacity absorbs growth you would otherwise hire for, value it at the cost of the hire you avoid, and say so. If it lets professional staff do more billable work, value it at the billable rate and state the assumed utilisation. If neither applies and the honest answer is reduced overtime and lower stress, use a lower rate and describe it as a capacity benefit rather than a cost saving. Reviewers respect a smaller honest number considerably more than a large unexplained one.

Use whatever rate your organisation applies to comparable investments. Your finance team will have one, and using it is better than picking a defensible-sounding number yourself. For Australian businesses this commonly falls somewhere between eight and fifteen per cent depending on sector, capital structure and risk appetite. Higher rates penalise benefits arriving later, which is appropriate for AI initiatives where the technology landscape may shift materially within the payback period.

It is the conventional choice and generally the right one, though for AI there is an argument for shorter. The technology is moving quickly enough that a capability built today may be commoditised within eighteen months, which cuts both ways. Your custom build may be superseded, but the underlying costs also tend to fall. If your case only works by counting benefits in year three, treat that as a warning sign. Strong AI initiatives generally pay back well inside two years.