The grade you leave the room with is not a single number on the final answer. It is dimension ratings, a short write-up, and a conversation among people who did not see your case. This is the firm side of that process: how forty-five minutes become a hire, a reject, or a "see them again."
I have sat in those decision meetings. The pattern that surprises candidates most is simple. A strong dimension almost never buys forgiveness for a broken one. Brilliant math with muddy structure reads as a risk. A correct close with weak communication reads as someone you cannot put in front of a client. Firms publish the dimensions; they do not publish the sheet. The process is still more structured than folklore admits.
After this, you should be able to read a firm's public language, imagine the note your interviewer will write, and stop optimizing for a composite score that does not exist.
What the firm is collecting (not "a score")
Consulting interviews are closer to a structured work sample than to a chat. Bain states on its hiring-process page that feedback rests on an "objective scoring system," and on Interviewing that interviewers use the same questions for a given role to reduce bias. Shared prompts, shared criteria, comparable notes.
McKinsey lists four themes for most client-facing interviews: personal impact, entrepreneurial drive, inclusive leadership, and problem solving (Interviewing). The case half is the problem-solving interview: structure ambiguous challenges, identify important issues, deal with facts and data, form conclusions and recommendations, and articulate your thoughts.
BCG's case interview preparation page covers the same family: structure, questions, data, calculations, prioritization, clear communication, intuition, creativity. Its interview preparation brochure (November 2018) adds prioritize, accurate analyses, synthesize, and creativity, and says there is no single correct answer; the interviewer wants a clear picture of thought process, skills, and potential.
None of those pages publishes a weighted 1–5 grid. The architecture is still clear: multiple dimensions, evidence in the hour, a judgment afterward. Your practice rubric can mirror the dimensions. It is not the firm's sheet.
Why structure exists (and what research says it buys)
Firms invest in shared dimensions and write-ups because unstructured interviews are noisy. Schmidt and Hunter's 1998 review in Psychological Bulletin put the average predictive validity of a structured interview at about .51 versus .38 for an unstructured interview for job performance. Later work debates exact magnitudes; the direction is stable: more structure, better prediction.
That is why your interviewer is not free to "just vibe" the hour. Dimension ratings force evidence into buckets. The write-up forces examples into sentences colleagues can challenge. The calibration meeting is where those sentences get stress-tested. Treat the case as a puzzle with a prize at the end and you optimize the wrong object. The firm is optimizing reliability across interviewers.
The three artifacts: ratings, write-up, meeting
In practice, the file that enters a decision room usually has three layers. Exact templates differ by firm and office. The shape does not.
| Artifact | What it holds | What it is for |
|---|---|---|
| Per-dimension ratings | Structure, analysis/math, judgment, communication, and (where relevant) behavioral themes | Makes strengths and holes visible without averaging them away |
| Interviewer write-up | Two to six sentences of evidence, often with a hire / no-hire lean | Gives colleagues something to argue with |
| Calibration / decision meeting | Multiple interviewers compare notes on the same candidate | Catches soft grades, inflates, and "but they were charming" |
I write the ratings while you talk or in the five minutes after you leave, and the note before I walk to the next room. If the note says "strong math, structure never landed, recommendation followed the last exhibit," the partner will ask for one example of each claim. Soft adjectives without examples get challenged. That is the point of the meeting.
"Your interviewer just decides if they like you" is half true: liking shows up on the communication and collaboration lines. It does not replace a missing structure line.
What MBB publish as the assessed dimensions
Treat the table as a translation of public language into sheet buckets, not as leaked weights.
| Firm (public source) | What they say they assess | How it shows up on a case sheet |
|---|---|---|
| McKinsey | Personal impact, entrepreneurial drive, inclusive leadership, problem solving; case tests structure, issues, data, conclusions, articulation | Behavioral themes on the PEI; problem-solving split across structure, analysis, synthesis, communication |
| BCG | Structure, questions, data, calculations, prioritization, communication, intuition, creativity; brochure adds prioritize, accurate analyses, synthesize | Structure and prioritization separate from analysis; creativity distinct from pure math |
| Bain | Logical, structured thinking; precision and creativity; clarify, structure, think aloud, listen and adapt; objective scoring | Structure and coachability visible; precision on math; creativity on options |
Two implications. Problem solving is not one dial. McKinsey's case description already lists five verbs; BCG's brochure lists four or five. Collapsing them into "case: 4/5" is a peer-mock habit. Interviewers do not do it. Behavioral themes are scored, not decorative. A failed McKinsey PEI on personal impact is not fixed by a clean case. A Bain conversation that never shows why you want Bain leaves a hole the system was designed to notice. Your own practice checklist belongs in a separate guide; this piece stays on the firm process.
How a strong dimension does and does not offset a weak one
The rule I have seen hold across Big Four and MBB decision rooms, without pretending to know any one firm's confidential matrix:
Offsets that sometimes work: one arithmetic slip you catch and correct; a thin industry fact if structure and logic still drive the recommendation; a slow start that recovers, if the write-up shows the recovery.
Offsets that almost never work: excellent math on a generic framework; a correct close reached by interviewer breadcrumbs; polish with no recommendation, or a recommendation that ignores the numbers; two interviewers marking the same dimension weak. One soft score is argueable. Two matching holes are a pattern.
Think in evidence, not averages. A 5 on math and a 2 on structure is not a 3.5. It is "can compute, cannot drive the problem." In a typical first-round slate of eight, my estimate is that two or three candidates have that profile: strong quantitative work, weak issue trees, shocked when they do not advance. The sheet already explained it.
Ned's rule. In the decision meeting, we hire patterns, not peaks. One brilliant dimension is a compliment in the write-up. Two weak dimensions with matching evidence across interviewers is the end of the file.
Worked example: two candidates, one sheet, one meeting
Original case, Harborline Apparel: mid-market outerwear, $420 million revenue, wholesale margin down from 38% to 32% in three years. Cut the mid-tier SKU line (about 28% of revenue) to protect margin? Exhibits on mix, returns, utilization. Two first-round candidates, same day. Ratings are Ned's 1–5 for illustration, not a published firm scale.
| Dimension | Candidate A (finance undergrad) | Candidate B (history major) |
|---|---|---|
| Structure / issue ID | 2 — opened with "revenue, cost, risk" then wandered | 4 — clarified goal (margin vs. growth), three drivers tied to SKU cut |
| Analysis / math | 5 — perfect contribution math, units narrated | 3 — one setup error, self-corrected, slower |
| Judgment / synthesis | 3 — recommended cut because "margin is down" | 4 — recommended phased cut; kept SKUs with high return rates last |
| Communication | 4 — clear, answer-first on math | 4 — clear, collaborative, adapted to a hint |
| Coachability | 3 — defended the generic tree when pushed | 5 — took the hint, rebuilt one branch in thirty seconds |
Write-up A (compressed): Strong quantitative execution. Structure never linked to the SKU decision; recommendation followed margin decline rather than drivers. Lean no.
Write-up B (compressed): Structure fit the decision; math adequate after self-correction; recommendation phased and risk-aware. Lean yes.
In calibration, someone will ask whether A's math should save them. No: structure and judgment are the day-one job. B's thinner math is discussable because the hole is small, self-caught, and surrounded by stronger evidence. A's peak is large and lonely. That is how offsetting works when people who hire argue a file.
Reproduce this: mid-tier contribution at 32% margin on 28% of $420m is 0.28 × 420 × 0.32 ≈ $38 million at stake. Candidate A recited it cleanly and still left a 2 on structure. The figure was never the grade.
What this means for how you prepare (firm side only)
Stop asking "did I get the case right?" Ask what evidence landed on each line of the sheet, and whether two interviewers would write the same weakness. Consistency across rounds beats one heroic close; the meeting compares notes, not vibes.
Audit with the firms' own verbs: McKinsey's structure / issues / data / conclusions / articulation; BCG's structure / prioritize / analyze / synthesize / communicate; Bain's clarify / structure / think aloud / listen and adapt. No evidence for a verb means that dimension is empty, whatever your final answer was.
CoachNed is independent of McKinsey, BCG, Bain, and every other firm named here; nothing above is an official scorecard.
Put one real sheet under pressure today
Run one timed case and force a firm-side debrief, not a vibe check. After the case, write four lines an interviewer would actually write: structure evidence, analysis evidence, synthesis evidence, one risk to the lean. Then score yourself 1–5 on each line and ask whether any peak is trying to excuse a hole.
The five-minute first rep on /start is enough to generate that note. For a full voice case with a seven-score debrief, use /interview (CopperCup, about eighteen minutes). Everything is open for seven days, no card; then $120 for a recruiting season or $49 a month.
Frequently asked questions
Do McKinsey, BCG, or Bain publish their exact scoring rubrics?
No. They publish dimensions and behaviors. McKinsey lists four themes and describes the problem-solving interview; BCG and Bain describe case behaviors; Bain also cites an objective scoring system. Exact weights and scales stay internal.
Can a perfect final answer save a weak case?
Rarely. Firms say they care how you approach the problem, not whether you land on one "right" answer. A correct close with incoherent structure still leaves thin evidence that you can drive a client problem without being led.
What happens in the interviewer decision meeting?
Interviewers bring ratings and short write-ups. The room looks for patterns across interviews: matching strengths, matching holes, and whether a soft grade has real examples behind it. Peaks get compliments. Repeated weak dimensions end the file.
Does strong math offset weak structure?
Usually not. Math is one dimension. Structure and judgment are closer to the day-one job. One arithmetic slip that you correct is forgivable. A brilliant calculation glued to a generic framework is not a hire case.
How is firm scoring different from a practice rubric?
A practice rubric trains you. Firm scoring feeds a write-up and a multi-interviewer decision. Put evidence on every public dimension; do not optimize for a composite average the decision room does not use.
Sources
- Interviewing | McKinsey & Company — four assessment themes and problem-solving interview behaviors. Checked 2026-09-24.
- Case Interview Preparation | BCG Careers — published case behaviors (structure, data, calculations, communication, intuition, creativity). Checked 2026-09-24.
- BCG Recruiting Interview Preparation Brochure (Nov 2018) — structure, prioritize, accurate analyses, synthesize, creativity; no single correct answer. Checked 2026-09-24.
- Our Hiring Process | Bain & Company — objective scoring system language. Checked 2026-09-24.
- Interviewing | Bain & Company — same questions per role to reduce bias; logical/structured thinking, precision, creativity. Checked 2026-09-24.
- Preparing for the Case Interview | Bain & Company — clarify, structure, think aloud, listen and adapt. Checked 2026-09-24.
- Schmidt & Hunter (1998), Psychological Bulletin — structured vs unstructured interview validity for job performance. Checked 2026-09-24.