A business case is usually commissioned by someone who already wants the answer to be yes. This is the ordinary situation, and it is the reason the method matters as much as the conclusion.
Agreeing the measure before measuring
The single most useful thing an assessment does is fix the definition of success before any figures are produced. Once a number exists, the definition tends to move towards it. Agreeing in advance which outcomes count, and in which of your existing reports they will be visible, avoids most of the disagreement that would otherwise arise at the end.
It also removes a common failure: the goal that cannot be measured at all. If nobody can say where success would show up, that is worth knowing in the first week of the work.
The full cost of ownership
Cases that overstate the return often understate the cost. The licence line is easy and everybody includes it. The lines that get left out are engineering time, the integration work that makes the tool usable in the workflow, the change management and the ongoing consumption that grows with use.
A case built on licence cost alone will clear almost any bar, and it is often revisited a year later by someone asking why the savings never appeared.
Writing a recommendation to stop
The document has to be readable by the person who approved the budget, and it has to withstand scrutiny from the team that favoured the investment. In practice that means three things: the arithmetic appears on the page itself, every assumption is listed with its source and the recommendation is stated in one sentence near the front.
It also means being specific about what would change the answer. A recommendation to stop is more useful when it says which condition, if it were different, would make the investment worth making. That turns a refusal into a decision the organization can revisit deliberately.
What it costs to be wrong in the other direction
An assessment that wrongly recommends against an investment has a cost too: the opportunity may not return, and the firm that produced the assessment rarely hears about it. The protection against both errors is the same: measured usage, full cost and an assumption list that somebody else can check.