Case mathematics
4 dimensions · 100 pointsAnswer
40
Working
25
Units
15
Interpretation
20
A correct figure delivered without its working cannot be checked, and one delivered without meaning leaves the interviewer to do the interpretation.
How your answers are scored
Six rubrics, four weighted dimensions each. The weights on this page come straight from the scoring code.
Why dimensions rather than one mark
A case is not one skill. Losing on synthesis calls for a different correction from losing on arithmetic, and a single score hides which one happened. Each rubric therefore reports its dimensions separately, and names the one to fix first.
Structuring
4 dimensions · 100 pointsCoverage
45
Separation
25
Depth
20
Concision
10
An interviewer follows a structure by ear. Coverage carries the most weight because a missing branch cannot be recovered later in the case.
Idea generation
4 dimensions · 100 pointsBreadth
30
Buckets
30
Specificity
20
Prioritisation
20
Breadth and grouping are weighted equally: a long undifferentiated list and three well-chosen angles both fall short on their own.
Market sizing
4 dimensions · 100 pointsScope
25
Assumption chain
30
Arithmetic
25
Sanity check
20
The estimate matters less than the chain that produced it. The sanity check is scored separately because it is what separates a defensible number from a guess.
Exhibit interpretation
4 dimensions · 100 pointsObservation
25
Implication
25
So what
30
Next test
20
Reading a figure is the cheapest part. The implication carries the most weight, because an exhibit is only worth what it changes.
Synthesis
4 dimensions · 100 pointsAnswer first
35
Evidence
25
Risk
20
Next step
20
The recommendation is tested on the opening sentence alone. A conclusion that arrives last arrives after the interviewer has stopped listening.
What the scoring does not do
Assessment is deterministic. The same answer always returns the same score and the same feedback, which makes results comparable across attempts. It also means the rubrics judge whether the expected moves are present, not whether a claim is true. A well-structured answer built on a false premise will score well here and fail in a room.
A human interviewer still reads what no rubric can: hesitation, the answer behind the answer, and whether you would be trusted in front of a client.
See it applied
Every rubric above runs on the free drills, without an account.