The recruiter email mentions a "case discussion" in your next round, and suddenly you are reading about MBA students who drilled two hundred practice cases, wondering how you are supposed to compete with that by Thursday. Take a breath. You are probably not interviewing at McKinsey. Case rounds for ops, strategy, and analyst roles at normal companies test something much smaller and much more learnable: can you take a messy business question and think through it out loud without falling apart. This guide is the honest version of what you can build in days, what you cannot, and why admitting uncertainty is your secret weapon.
What a case interview is actually testing
A case is a business problem posed live: "Our warehouse costs rose 20 percent, what would you look at?" or "Should this retailer open a second location?" You will not have enough information, on purpose. The interviewer wants to watch your process, not grade your final answer.
Across companies, the scoring usually reduces to four things:
- Structure: do you break the problem into parts before diving in?
- Reasoning out loud: can the interviewer follow your thinking in real time?
- Numeracy: when arithmetic appears, is it organized and roughly right?
- Judgment: do you distinguish what you know from what you are assuming?
Notice what is not on the list: memorized frameworks, industry expertise, consulting vocabulary. Outside strategy firms, interviewers do not expect any of that, and reciting a canned framework that does not fit the question actively hurts you.
What is realistically learnable in one week
Be honest with yourself about scope, because trying to learn everything guarantees learning nothing. In a week you can genuinely build three skills.
1. The structure-first habit (days 1 and 2). When the question lands, do not answer. Say: "Let me take a moment to structure this." Then split the problem into two to four pieces that together cover it, say them out loud, and ask which to start with. That is it. That single habit separates you from most unprepared candidates.
You do not need branded frameworks; you need the splitting move behind them. Profit is revenue minus cost. Revenue is price times volume. Costs are fixed or variable. A process has steps, and the problem lives in one of them. Practice by structuring ordinary questions aloud: why is this coffee shop losing money, should our family buy or lease a car. Ten of those and the habit starts to stick.
2. Math hygiene (days 3 and 4). Case math is arithmetic under observation, which is a different sport from arithmetic. The failures are almost never ability; they are dropped zeros and lost tracks. The fixes are mechanical:
- Write everything down, and narrate as you go: "So 2 million customers times 5 dollars is 10 million."
- Round aggressively and say so: "I will call 48 roughly 50 to keep this clean."
- Sanity-check the result against reality before presenting it. If your answer implies every American buys nine mattresses a year, say that it smells wrong. Catching your own absurd number scores points; missing it costs them.
Practice market-sizing questions ("how many gas stations in Texas?") because they exercise both estimation and hygiene at once.
3. Full cases out loud (days 5 to 7). Reading cases silently does almost nothing; the skill is verbal. Grab free case prompts online, set a timer for 25 minutes, and talk through them alone or with a friend playing interviewer. Three to five full reps is enough to stop the format from feeling alien, and that is the real goal of a one-week prep: not mastery, familiarity.
What is not learnable in a week (and why that is fine)
Speed, polish, and deep pattern recognition come from hundreds of reps you do not have time for. Do not fake them. A candidate performing consulting polish they do not have is easy to spot and hard to trust, and interviewers at non-consulting companies are not grading on polish anyway.
You also cannot cram business intuition about industries you have never touched. If the case is about pricing airline seats and you have never worked near airlines, your permission slip is one sentence: "I have not worked in this industry, so tell me if my assumptions are off." Then reason from common sense. Interviewers give the data when asked; asking good questions is scored behavior, not weakness. If you are also facing a skills-based round you have not used in years, the triage mindset in technical interview prep when you are rusty pairs well with this plan.
Why a flagged assumption beats a bluffed fact
Here is the moment that decides most beginner cases. You need a number you do not know, say, average customer spend. You have two options:
- Bluff: state "average spend is 40 dollars" like a fact and keep moving.
- Flag: say "I do not know average spend. I will assume 40 dollars based on my own grocery runs, and this is worth checking, because the whole answer moves with it."
The bluff feels stronger and scores worse. Interviewers probe: "Where did 40 come from?" The bluffer either doubles down on a fiction or visibly deflates. The flagger has already answered the probe before it was asked, and has demonstrated the exact judgment the interviewer is hired to detect: knowing the difference between what you know and what you assume. In real analyst and ops jobs, that difference is the whole job. There is a broader version of this skill in what to say when you do not know the answer.
The same rule applies to your conclusion. End with a recommendation, state it plainly, and attach its biggest risk: "I would test the price increase in two regions first, because my demand assumption is the weakest link." Decisive and honest at the same time; that is the winning register.
Your resume is part of the case
Case rounds rarely float free of your history. Interviewers open with "walk me through the analysis you are most proud of," or close by picking a bullet off your resume, usually one with a number in it, and probing: how was that 15 percent measured, what was your part versus the team's, what would you do differently. In effect, your own resume becomes a case, and you are the dataset.
That is worth preparing deliberately. Reread every quantified claim on your resume and rebuild each number from memory: source, method, your role. If you cannot, reword the bullet to what you can defend, because a bullet you cannot reconstruct under probing does more damage than a modest one.
The free Bounce scan at careerbounce.io helps you do this audit properly. It shows you exactly what the employer's software parsed from your resume, every claim and keyword as the interviewer will see it, so nothing on that page surprises you in the room. It is free, it runs entirely on your device, and your resume never leaves your computer. No scan can pass the case for you. But walking in with a structure habit, clean math, flagged assumptions, and a resume you can defend line by line is a genuinely competitive package, consulting background or not.