How Case Interviews Are Scored: What Interviewers Actually Grade

You walk out of a case interview with a feeling; the interviewer walks out with a form. Firms keep their exact rubrics internal and they vary in wording, but across firms' publicly described criteria and two decades of prep-community consensus, evaluations converge on the same four dimensions. Knowing them changes how you prep — because you stop practicing "cases" in general and start improving the specific line item that's dragging your score.

The four dimensions

1. Structure. Can you turn an ambiguous prompt into an organized, case-specific plan of attack — and then actually use it to navigate? Graded at the opening framework, yes, but also every time you organize a brainstorm or close a branch. Recited frameworks score low precisely because this dimension is testing organization of thought, not recall (why memorized frameworks backfire).

2. Quantitative reasoning. Setting up the right calculation, executing it accurately while narrating, and — the part candidates miss — saying what the number means. A correct figure left uninterpreted scores like a chart left undescribed. Appetite counts here too: volunteering for the math reads differently than being dragged to it (the case math skill).

3. Business judgment. The instinct layer: are your assumptions sane, do your hypotheses sound like someone who understands how businesses make money, does your recommendation weigh the risk a real client would care about? This is the dimension career switchers most often need to build — and it grows from exposure to many cases across industries more than from any drill.

4. Communication. Not polish for its own sake — followability. Top-down answers, signposted transitions, math narrated as it happens, synthesis on request. Communication is the multiplier dimension: it's how the other three become visible. An interviewer can't credit a structure they couldn't follow.

Alongside the four, interviewers typically record an overall hire/no-hire recommendation — and in practice a serious weakness in any single dimension is hard to average away. The bar is no-major-weakness, not one-great-strength.

Why you can't diagnose yourself

Self-assessment after a case is notoriously unreliable, in a predictable direction: candidates overweight the math (it's the part with a right answer, so errors feel catastrophic) and underweight communication (you can't hear your own followability). The result is months spent drilling arithmetic while the actual flag was "hard to follow" — a completely different fix. This is the argument for making a meaningful share of your practice scored practice: mocks with a coach who grades dimensions, or AI mocks with a structured breakdown. In Key Takeaways, every session ends with a scored breakdown skill by skill, from structuring and quant to communication, so after a handful of cases you're looking at a pattern, not a feeling.

Turning your weak dimension into a prep plan

The whole point of knowing the rubric is targeting:

• Structure low → daily 90-second structure sprints on fresh prompts, plus interrogating every bucket ("why is this on the list?").
• Quant low → the daily narrated-math block; volunteer for every calculation in mocks even when it's optional.
• Judgment low → widen case exposure across industries; after each case, ask "what would a partner have worried about that I didn't?" and read business press with that lens.
• Communication low → record your mocks and listen for buried leads; practice answer-first delivery ("I recommend X, for three reasons...") until top-down is your default shape.

Re-test in scored mocks after each targeted week. A dimension usually moves within two or three weeks of aimed practice — but only if the aim was right, which is the diagnosis's job.

The bottom line

Four dimensions, graded together, with no-major-weakness as the bar. Stop practicing generically: find out which line item is costing you — from a coach's grades or an AI mock's breakdown — aim a week of drills at it, and re-measure. Prepping against the rubric is the closest thing case prep has to a cheat code, and it's fully legal.