Ten structural questions that reveal whether a business case is built to be examined or built to be approved, answered from your knowledge of the case with nothing uploaded.
You do not upload the business case and no document is read or analysed by anything. You answer ten questions about how the case is constructed, and the fixed-point scoring shows where its structure would give way under an approver's pressure.
Last reviewed 3 July 2026 · Free and ungated
Put the case to selected senior operators who have delivered comparable commitments and receive confidential challenge on the assumptions that matter.
Pressure-test this decisionAnswer from your knowledge of the case as written today. No document is uploaded or analysed.
As described, the case is a recommendation wearing the clothes of an analysis. Without stated assumptions, compared alternatives or a genuine downside, approvers are being asked to endorse a conclusion rather than examine a case.
The skeleton of a sound case is present, but specific joints are weak: typically benefits nobody has signed up to deliver, timelines set by ambition, or validation that never left the building. These are the elements approvers probe first, and the ones that unravel first after approval.
The case is structurally complete: assumptions, alternatives, downside and ownership are all visible. What structure cannot guarantee is honesty of inputs. A well-built case can still rest on a benefits figure that was reverse-engineered from the approval threshold.
This stress test never sees your business case. There is no upload, no text analysis and no AI reading anything: you answer ten questions from your own knowledge of the document, and fixed point values do the rest in your browser. That constraint is deliberate: the structural weaknesses that sink cases are ones the author already knows about when asked directly.
The ten questions cover the load-bearing elements of any investment case: whether the assumptions are visible enough to be challenged, whether alternatives were genuinely developed, whether anyone owns the benefits, and whether the approvers know what would make them say no. A case can have impeccable arithmetic and still fail all ten, because arithmetic is the easy part.
Each answer carries 0, 5 or 10 points, summed and normalised to 100, then mapped to one of three structural bands. Because the same answers always give the same score, the most useful pattern is the retest: strengthen the two weakest answers, run it again, and repeat until the structure stops being the problem.
Because the cases that matter are confidential, and because structural weakness is not a text-analysis problem. The author already knows whether anyone owns the benefits; the test simply forces that knowledge into the open without any document leaving your machine.
Structure gets a case examined fairly; it does not make the underlying investment good. Rejection of a well-structured case usually means the substance was tested and found wanting, which is the process working as intended.
Both, separately. The gap between the author's score and a reviewer's score is one of the most revealing outputs: it shows where the case reads as complete to the person who wrote it and incomplete to everyone else.