By minute ten I can usually tell which candidates have been casing with friends: fluent, signposted, and lost the first time I say "I don't think that's right." The candidate half of their mock sessions was fine. The interviewer half was two people taking turns at a job neither had learned, and a practice partner is worth exactly as much as the interviewer they can be for you.
Most advice on partners covers where to find one and how to perform as the candidate. Almost none covers the other chair: reading a prompt without helping, staying silent, disagreeing on purpose, writing down what was actually said, and scoring the case rather than the person. That is what you need from a partner, and what you owe them back. Below: how to find one, and how to do the job for them.
What the interviewer is doing while you talk
In a real first round I sit down with an answer key, a clock, a pen, and a plan. The plan includes one moment where I will disagree with you whether or not you are right, because the sheet has a line for how you take it. A peer interviewer usually has none of the four. They read the prompt, become an audience, and say "nice" at the end.
What the sheet rewards is mostly process. The recommendation is one line out of six or seven; the others ask whether the structure could organize a real team, whether the math was set up before it was done, whether the close led with the answer. A partner who is not watching for those things cannot score them, and a partner who cannot score them is a friendly clock.
The firms say this in public. McKinsey states that most client-facing candidates get a personal experience interview followed by a problem-solving interview, and that its sample cases "contain logic, thought processes, and suggested answers." BCG's preparation page says "what matters most is how you approach the problem and the quality of your reasoning." Bain's Coffee Shop Co. sample prints "a good answer" and "a better answer" side by side. All three are written for the candidate; the person who needs to have read them is the one about to interview you.
Where partners come from, and how many there are
Density decides your strategy. At Kellogg, consulting took 38% of the class of 2025's accepted offers. At Harvard Business School, 21% of graduates who took jobs went into consulting. Where a third of your section is casing, your problem is quality control; where nobody is, it is finding one person.
| Source | What you get | What goes wrong | Best use |
|---|---|---|---|
| Consulting club or MBA section | Volume, easy scheduling | Politeness; the same three frameworks | Weekly reps; rotate partners every three sessions |
| Recent hires and alumni | Someone who sat on the other side this year | Scarce; one hour, not ten | A checkpoint two weeks out |
| Online matching communities | A partner at 9 p.m. on a Tuesday | No-shows; people who only want to be the candidate | Filling gaps, with a hard screen |
| A friend outside consulting | Honesty and availability | They do not know the sheet | Communication and synthesis, card in hand |
The ask is four lines: what you are preparing for and when, the format and length you want, the cadence you can commit to, and the offer to interview them in return with the key read the night before. Almost nobody makes the fourth offer, which is the line that gets a yes.
The first session is the interviewer's audition, not yours. Did they read the key beforehand, write anything down while you spoke, push back at least once, and quote a sentence you actually said? Two of four, book again. Fewer, do not rebook.
The interviewer's card: nine moments in 25 minutes
Print this and keep it beside the case. You cannot quote what you did not write down.
| Clock | You do | You watch for | You write down |
|---|---|---|---|
| 0:00 | Read the prompt once; repeat only if asked | Do they note the numbers? One clarifying question or five? | Their restatement, verbatim |
| 0:02 | Say "take your time," then nothing | Under 30 seconds usually means a memorized framework | Seconds of silence |
| 0:04 | Listen to the structure without nodding | Overlapping buckets; can you tell where they would start? | Their bucket names |
| 0:07 | Release data only when asked, plainly | Do they say what they expect before hearing the number? | What they asked for, in order |
| 0:11 | Hand over the exhibit; stay quiet for twenty seconds | Is the first sentence the title or the insight? | Their first sentence |
| 0:14 | Pose the math; ask for the approach before the arithmetic | Approach, units, sanity check | The answer and how long it took |
| 0:18 | Planned pushback: disagree once, right or wrong | Fold, defend with data, or ask a question back | What they said next |
| 0:21 | "What else could be driving this?" | A list, or categories? | Idea count |
| 0:23 | "The CEO has sixty seconds. What do you tell her?" | Answer first, then reasons, risk, next step | Their opening sentence |
At 0:25, stop, and both of you score in silence. Two habits carry most of the value. Silence: saying nothing while a friend struggles is the hardest thing a peer interviewer does; the real one does it every time. And the planned pushback: pick the moment in advance and disagree there regardless of the answer; the sheet is scoring the candidate's next two seconds, not whether you were right. To see the rhythm, watch Bain's Consultant mock interview as the interviewer, noting when the Bain interviewer speaks and for how long they do not.
Run this case tonight: Bramblewood Pet Clinics
The interviewer reads this the night before; the candidate does not.
Prompt. "Bramblewood operates 40 veterinary clinics across the US Midwest. Its private equity owner wants $5 million more in annual profit within two years without buying clinics, and asks whether extending weekday opening from 6 p.m. to 8 p.m. is the way to get it. What do you think?"
Data, released only when asked. A clinic sees 9,000 visits a year, open 8 a.m. to 6 p.m. on weekdays for 50 weeks. Revenue per visit averages $200; variable cost per visit is $80. Evening staffing costs $150,000 per clinic per year.
Math question. How many extra visits does one clinic need to break even on the evening hours, and is that realistic?
Key. Contribution per visit is $200 minus $80, or $120. Break-even is $150,000 divided by $120: 1,250 visits a year, about 25 a week, 2.5 an hour across ten evening hours. Daytime runs 3.6 an hour (180 visits over 50 open hours), so evenings need roughly 70% of daytime throughput. Plausible. The $5 million target needs $125,000 of profit per clinic: ($150,000 plus $125,000) divided by $120 is about 2,300 visits a year, 46 a week, 4.6 an hour. That is a busier clinic at 7 p.m. than at 11 a.m., and the candidate who notices has found the case.
Planned pushback, right after the math. "Our regional managers say vets refuse evening shifts at any price." A strong answer asks whether that means at any price or at current pay, then reprices: a 20% premium lifts fixed cost to $180,000 and break-even to 1,500 visits, 3 an hour. Still under daytime.
Synthesis prompt. "The owner's partner is in the lobby. Sixty seconds." Key answer: evenings can pay for themselves but cannot deliver $5 million alone; pilot five clinics.
Two deliveries of the same correct number, scored 1 to 4 on three lines of the sheet:
| What the candidate said | Math | Communication | Judgment |
|---|---|---|---|
| "So, 150 over... 200 minus 80 is 120... 150 over 120 is 1.25, so 1,250? Yeah. 1,250 visits." | 4 | 2 | 1 |
| "Break-even is fixed cost over contribution: $150,000 over $120 is 1,250 visits a year, 25 a week, 2.5 an hour against 3.6 in the daytime. Plausible. But the $5 million target needs 46 a week, busier than the daytime clinic." | 4 | 4 | 4 |
Same number, roughly twice the points. The arithmetic is one line on my sheet. The road to it is three.
Score in silence, then talk in quotes
Swapping roles creates a debt: each of you owes the other a kind review, and a real interviewer owes you nothing. The fix is mechanical. When the case stops, both fill in the same sheet for three minutes in silence; the candidate scores themself. The interviewer then reads three verbatim quotes (one that earned a point, one that lost one, one ambiguous) and asks about the ambiguous one before offering a view. Only then compare sheets. The largest gap is tomorrow's drill, written down before you swap.
| Dimension | Sounds like a 2 | Sounds like a 4 |
|---|---|---|
| Structure | Overlapping buckets, such as "external factors" holding competitors and regulation | Buckets a team could split across, with a stated first branch |
| Prioritization | Works every branch in order | Names the branch that moves the answer most and asks for that data first |
| Math | Right number, no approach, units on request | Approach aloud, then arithmetic, then a sanity check |
| Exhibits | Reads the title and the axes | Insight first, then what it means for the client |
| Business judgment | "It depends" | A view, plus the one assumption that would change it |
| Synthesis | Recap of everything, answer at the end | Answer first, two reasons, one risk, one next step, under a minute |
| Communication | You have to ask where this is going | Signposts each move; stops when the point is made |
Quoting rather than characterizing is not politeness. Kluger and DeNisi's 1996 meta-analysis of 607 feedback effects found that feedback improved performance on average but that over a third of interventions made it worse, with effectiveness falling as attention moved from the task toward the self. "You seemed nervous" is about the self. "Your second bucket held both competitors and regulation, which overlap your market bucket" is about the task. Only one of those can be fixed by Thursday.
Ned's rule. Feedback that does not quote you is a mood report. Make your partner write down three things you said, verbatim, before they tell you what they thought of them.
When the partner is the wrong tool
Some problems a partner cannot fix; some sessions are telling you to change the partner.
| Symptom | What is actually wrong | Fix |
|---|---|---|
| The same note three sessions running | You are rehearsing a leak, not repairing it | Pause mocks; drill that dimension alone for a week |
| Your partner hands you the framework you "should" have used | They are casing you as a peer, not interviewing | Give them the card; if it persists, swap |
| Every session ends "that was solid" | Reciprocity | Silent scoring first; if the sheets never differ, new partner |
| You freeze at pushback | Too little of it | Two pushbacks per case, plus a voice case that pushes back |
On cadence, my estimate from watching what works: one partner session a week from six weeks out, two a week in the final fortnight, none in the last three days.
Practice this today
Tonight, send the four-line ask to two people. Before the session, sit the five-minute first rep at /start: three typed turns of a case, instant scores, no account, so you know your weakest line. Then run Bramblewood with the card, and swap.
If nobody is free this week, run the live voice case at /interview: UrbanBrew, about 18 minutes, a seven-score debrief, and it pushes back. The case library has 64 original cases with the interviewer's data written out; the synthesis drill is the sixty-second close on its own.
Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month. CoachNed is independent and not affiliated with McKinsey, BCG, Bain, or any other firm named here.
Frequently asked questions
How do I find a case interview practice partner?
Start with the densest pool you have: a consulting club or MBA section, then alumni and recent hires, then online matching communities with a hard screen. Offer to interview them in return with the key read in advance.
Should my case practice partner be better than me?
Slightly better, or simply stricter. Two beginners work if they use a written sheet, score in silence, and quote each other.
Can I prepare for case interviews without a partner?
Mostly, yes. Structure, math, exhibits, and synthesis drill well alone or with software. You need a human for pressure: interruption, disagreement, and the follow-up that wanders.
Sources
- Interviewing at McKinsey — interview format; sample cases include logic and suggested answers. Checked 2026-09-24.
- BCG case interview preparation — "how you approach the problem and the quality of your reasoning." Checked 2026-09-24.
- Bain, Coffee Shop Co. sample case — "a good answer" and "a better answer" side by side. Checked 2026-09-24.
- Bain, Consultant mock interview — official mock case video. Checked 2026-09-24.
- Kellogg Full-Time MBA career outcomes, class of 2025 — consulting 38% of accepted offers; median base $190,000. Checked 2026-09-24.
- HBS employment data, consulting, class of 2025 — 21% went into consulting; median base $190,000. Checked 2026-09-24.
- Kluger and DeNisi (1996), feedback interventions meta-analysis — 607 effect sizes; d = .41; over one-third decreased performance. Checked 2026-09-24.