Most mock feedback dies as a sentence. "Be more structured." "Own the math." "Close stronger." You nod, book another case, and leak the same points. The firms already publish what they listen for. Your job is not another tip sheet. It is a printable score sheet you fill in yourself before anyone else grades you.
After this piece you should be able to run a fifteen-minute pre-interview checklist, score a mock on six rows in under five minutes, and turn the lowest row into one drill for the next day. How firms mark internally is a different topic; this page stays on your checklist and your rubric.
What McKinsey, BCG, and Bain say they assess
None of the three firms posts a confidential scoring grid. They do publish, on their careers pages, the skills the case is meant to surface. Map those phrases to practice rows and you stop inventing criteria.
| Firm (public page) | What they say they evaluate | Rubric row it maps to |
|---|---|---|
| McKinsey interviewing | Structure ambiguous challenges, identify important issues, deal with facts and data, formulate conclusions, articulate thoughts | Structure, data handling, synthesis, communication |
| BCG case prep | Structure the approach, ask thoughtful questions, analyze data, calculate quickly, spot key factors; clear logic, business intuition, creativity | Structure, inquiry, quant, judgment, communication |
| Bain case prep | How you think and structure; clarify the objective; walk through a framework; think aloud; listen and adapt | Structure, clarifying, transparency, adaptability |
| Bain recruiting advice (Mar 2026) | Adapt the approach to the problem; take a reasoned position; quality of reflection over case volume | Hypothesis / judgment, prioritization |
McKinsey frames the case as a problem-solving interview on a client scenario. BCG states there is not always a single right answer and that quality of reasoning matters most. Bain tells candidates to clarify first, structure out loud, and pause when needed. Treat those as the design brief for your self-score, not as folklore from forums.
Printable self-scoring rubric
Print this table. Score each row 1–5 within five minutes of ending the case, before you talk yourself into a better performance. Use half-points only if you must. Total is out of 30.
| Dimension (1–5) | 1 — fail | 3 — passable | 5 — interview-ready | Your score |
|---|---|---|---|---|
| Structure | Generic buckets; overlap; no reason to start anywhere | Case-specific tree; one overlap or weak MECE | Clean tree; each branch tied to the decision; clear first dive | |
| Clarifying and focus | No clarifiers, or trivia; drifts off the objective | 1–2 useful clarifiers; mostly on-objective | Clarifiers that change scope or metric; stays on the ask | |
| Quant process | Starts calculating cold; units wrong; no check | Formula named; one unit or rounding slip | Setup spoken; units locked; intentional rounding; sanity check | |
| Data and exhibits | Misreads chart; ignores units or axes | Correct read; slow insight | Reads labels first; extracts the decision-relevant number | |
| Judgment and hypothesis | Lists options; never bets; follows interviewer only | States a soft bet late | Early, revisable hypothesis; prioritizes the highest-leverage branch | |
| Synthesis and communication | Trails off; no recommendation; mumbles transitions | Recommendation present; thin risks or next steps | Answer-first close; 1–2 supports; risk + next step; clear signposts |
How to use the numbers. A 5 means you would not change that moment under first-round pressure. A 3 means the interviewer would not fail you on that row alone but would not champion you either. A 1 means a partner would stop writing and start coaching.
Ned's rule. Score yourself before you open anyone else's feedback. If your lowest row and theirs disagree, trust the lower of the two until the next mock proves otherwise. Optimism is not a dimension on the sheet.
Pre-interview checklist (print, check off)
Run this the night before a real interview or fifteen minutes before a mock. It is logistics plus one cognitive goal. Do not turn it into a second case.
| When | Item | Done |
|---|---|---|
| Night before | Confirm format (candidate-led vs interviewer-led), length, video vs in-person | |
| Night before | One PEI or fit story with a number and a decision you owned | |
| Night before | Pen, blank paper (or tablet), water; for video: lighting, mute notifications, test mic | |
| 15 min before | Pick one rubric row to protect this hour (not all six) | |
| 15 min before | Write the decision question in one sentence; leave space under it for clarifiers | |
| Door / join | Restate the objective aloud before structuring; ask for units, time horizon, and success metric if missing | |
| First 3 min | Pause 60–90 seconds; state structure; propose where you start and why |
Bain's own tips open with "clarify the objective" and "structure your thinking." That is not optional polish. If those two boxes are empty, the rest of the checklist does not matter.
Worked example: score one close in ninety seconds
Invented client: Northline Transit, a regional bus operator deciding whether to raise off-peak fares 12% to close a $4.8M annual gap. You have already found that off-peak riders are 38% of trips and fare-elastic at roughly −0.6.
Weak close (score synthesis 2): "So fares are sensitive… maybe we should look at costs too… I think a smaller increase could work, and we should talk to marketing."
Missing: answer-first position, math link to the $4.8M gap, risk, next step.
Stronger close (score synthesis 5): "Raise off-peak 8%, not 12%. At −0.6 elasticity, a 12% hike cuts off-peak volume ~7%, so net revenue lift is closer to 4–5% of off-peak fare revenue — short of $4.8M if off-peak is under half of fare income. An 8% hike with a timed trial on two corridors, plus a parallel cost cut of $2M in empty peak trips, closes the gap with less churn risk. Next step: four-week A/B on corridors A and C; kill if ridership falls more than 5%."
Reproduce the elasticity step: percent volume change ≈ −0.6 × 0.12 = −0.072 (−7.2%). Net price effect on that segment ≈ 1.12 × 0.928 − 1 ≈ +4%. That is the number the recommendation hangs on. On the rubric you also mark Quant 5 (setup + check) and Judgment 4–5 (chose a smaller hike plus a cost lever instead of forcing the client's preferred 12%).
A peer who only says "nice close" leaves you with nothing to drill. Your own scores name the row.
Post-mock review protocol (ten minutes, not thirty)
Do this the same day. Delay is how "be more structured" returns.
- Fill the rubric cold (3 min). No rewatching yet. Circle the lowest row.
- Write one evidence line (2 min). Quote yourself, or note the timestamp: "At minute 6 I opened with Cost / Revenue / Risk with no link to the fare decision."
- Name one drill (1 min). Map the lowest row: Structure → structure drill; Quant → math; Exhibits → charts; Synthesis → synthesis; Hypothesis/judgment → brainstorming under a hypothesis prompt.
- Schedule the next rep (1 min). Same case type within 48 hours, or a full voice case if the leak only shows under pressure.
- Only then open a partner's notes, a coach, or an AI debrief (3 min). Compare their lowest theme to your lowest row. If they praise a row you scored 2, keep your 2.
CoachNed's live voice case at /interview (CopperCup, about 18 minutes) ends in a seven-score debrief: Structuring, Hypothesis, Math, Charts/exhibits, Synthesis, Communication, and Business judgment. Use it as a second opinion after your six-row sheet, not as a substitute for filling the sheet yourself. The seven scores are finer-grained; your printable table is what you can run with a peer who will not grade you.
Weekly use without turning prep into bureaucracy
Three mocks a week is enough for most candidates in the eight weeks before first rounds (Ned's estimate, not a study). After each mock, only the lowest rubric row earns drill time that day. Cap drills at twenty-five minutes. If Structure is a 2 three mocks in a row, you have a stable weakness; switch case types and keep scoring Structure until it is a 4. If Structure flips between 2 and 5, your problem is consistency under interruption, not ignorance of frameworks — practice with a partner who cuts in, or the voice room.
Do not average scores across the week into a vanity total. Track the floor of each row. Interviewers remember the softest moment, not your mean.
Practice this today
- Print the rubric table above.
- Run one full case out loud — peer, club, or the /interview voice case.
- Score all six rows before you read any debrief.
- Spend twenty minutes on the drill that matches your lowest row: /drills/structure, /drills/math, /drills/charts, /drills/synthesis, or /drills/brainstorming.
Five minutes and no account: /start (three typed turns, instant scores). Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month.
Frequently asked questions
Should I use the same rubric for McKinsey and Bain?
Yes for self-scoring. The public skill lists overlap heavily: structure, data, judgment, clear communication, and a recommendation. Format differs (more interviewer-led at McKinsey in many offices; more candidate-led at Bain and BCG), but the rows stay the same. Adjust only the Clarifying and Focus row: interviewer-led cases reward crisp answers to prompts; candidate-led cases punish a missing first dive.
How many mocks before I trust my self-scores?
After roughly eight to ten scored mocks, most candidates' self-scores stabilize within one point of a tough partner (Ned's estimate). Until then, treat your scores as a bias detector: if you always give yourself a 4 on communication, ask a partner to score that row only.
What if my partner and I disagree by two points?
Take the lower score, write one evidence line each, and re-run the same opening or close for five minutes. Disagreement without evidence is noise. Disagreement with two quotes is useful.
Can I skip the checklist on interview day?
Skip the night-before logistics only if you have already packed. Do not skip restating the objective and pausing to structure. Bain and BCG both put structure and clear reasoning at the center of the exercise; starting cold is how strong analysts look average.
Is an AI debrief enough without a human partner?
It is enough to move a specific row week to week. It is not enough for presence, interruption handling, and the social texture of a partner interview. Alternate: scored solo or AI reps for volume, human mocks for interruption.
Sources
- Interviewing at McKinsey — problem-solving interview skills: structure, issues, data, conclusions, articulation. Checked 2026-09-24.
- BCG case interview preparation — step-by-step case expectations; reasoning over a single right answer. Checked 2026-09-24.
- BCG interview process — case assesses problem-solving and analytical skills alongside collaboration cues. Checked 2026-09-24.
- Preparing for the Bain case interview — clarify, structure, think aloud, listen and adapt. Checked 2026-09-24.
- How to stand out in Bain's consulting recruiting process — tailor the approach; quality over volume; reasoned position (Mar 2026). Checked 2026-09-24.
CoachNed is not affiliated with McKinsey, BCG, Bain, or any firm named above.