Case Interview Practice Partner: Script, Scorecard and Schedule

Why practice case interviews with a partner?

A partner turns case knowledge into interview performance. Reading solved cases teaches patterns. A live counterpart adds the pressure to ask for data, calculate aloud, recover from a wrong turn and close under time pressure. Those are the behaviors interviewers score.

The best partner sessions follow a script, use a fixed scorecard and end with one written fix. Unstructured sessions tend to slide into discussing the case content, which feels productive and changes little. The rest of this guide gives you the script, the scorecard, a group format, a schedule and a case to run tonight.

Where to find a case partner

Most candidates combine three or four sources. Each brings a different kind of feedback.

How to give a case: an interviewer script

The interviewer's job is to simulate a real interviewer: neutral, prepared and slightly demanding. Read the whole case beforehand, including the numbers, so you can check the math in real time.

Four reveal rules keep the session realistic:

1. Give data only in answer to a specific request. A vague question such as "anything else on costs" gets "what would you like to know".
2. Hint once, after 30 seconds of being stuck, with a direction only ("look at it per bike").
3. Let wrong math run to the interpretation stage, then ask the candidate to sanity-check. Real interviewers often do the same.
4. Stay neutral in tone. Nodding at every step trains a candidate to look for approval.

The feedback scorecard

Score each dimension 1-5 using the anchors below; 2 and 4 sit between them. Keep the scores in a shared sheet so you can see trends across partners. The six dimensions mirror what firms say they assess: problem solving in the case and personal qualities in the fit part.

The case interview recommendation guide gives a second-by-second template for the synthesis row, and the list of case interview mistakes helps with the 1-level anchors.

Practicing in a group: the three-role format

Three people make the most efficient practice unit: one candidate, one interviewer and one observer. The observer keeps time, fills in the scorecard and notes exact quotes, which lets the interviewer concentrate on running the case.

Rotate roles each round, so a 2.5-hour session gives everyone one case in each role. Observing is surprisingly valuable, because watching a peer calculate in silence or close without a recommendation makes the same habit easy to spot in yourself. For assessment-center formats with several candidates in one case, use the separate drills in the group case interview guide.

How many practice cases are enough?

Most candidates need 30-50 live cases over four to six weeks. No firm publishes a number, and the range is a planning estimate; candidates with strong math or prior business training often need fewer.

Research supports the focus on feedback over volume. The deliberate-practice concept, from Ericsson, Krampe and Tesch-Römer in 1993, defines practice as effortful work on specific weaknesses with feedback. A 2014 meta-analysis by Macnamara, Hambrick and Oswald found that practice explained about 26% of performance differences in games, 21% in music and 18% in sports, and less than 1% in professions. Case performance depends on much more than repetitions, so each session should target one logged weakness.

A six-week partner schedule

The schedule fits the six-week preparation plan. Candidates with four weeks can merge weeks one and two and weeks four and five.

That totals 32 cases as candidate. Draw prompts from the list of 20 original practice cases and rotate partners every week or two so no one memorizes your style.

A ready-to-run practice case

Prompt (read aloud): Our client runs 2,000 shared e-bikes in Utrecht. Revenue has stayed flat for two years, but annual profit fell from EUR 1.2 million to EUR 0.4 million. The CEO wants to expand to a second city next year and asks whether she should.

Clarifying facts (give only if asked): The fleet size has not changed. The goal is to restore 2024 profit before any expansion. Prices rose in 2025.

Exhibit (give when the candidate asks about revenue or costs):

Calculation 1: Revenue per bike per year. 2024: 4.0 × 365 × 2.50 = EUR 3,650. 2026: 3.2 × 365 × 3.10 = about EUR 3,620. Revenue per bike is roughly flat.

Calculation 2: Profit per bike. Costs rose from EUR 3,050 to EUR 3,420. Profit per bike fell from EUR 600 to about EUR 200, which times 2,000 bikes matches the EUR 1.2 million and EUR 0.4 million. Of the EUR 400 drop, about EUR 370 comes from costs: EUR 220 from theft and vandalism and EUR 150 from maintenance.

Brainstorm question: "What could explain the rise in theft and vandalism?" Strong answers group ideas into bike design (locks, tracking), parking (free-floating versus docked zones) and user behavior (account checks, deposits).

Model close: "She should postpone the second city. Profit per bike fell by two-thirds, and over 90% of that comes from higher costs, mainly theft, vandalism and the repairs that follow. Expanding now would copy a loss-making cost base. The risk is that ride volume, down 20% since the price rise, keeps falling. Next steps: test tracked locks and fixed parking zones in two districts, and measure price sensitivity before the next price change."

Partners who want more cases at this level can use the three fully worked candidate cases in Stephan's book Cracking the Case, written so one person can interview the other.