The fastest way to learn case interviews is not more reading. It is a scored spoken rep, one specific fix, then another spoken rep. Candidates with an interview in two weeks who bury themselves in guides feel busy and still freeze on the first clarifying question. After this piece you will have a 14-day calendar with hour counts, a clear stop-list for what cannot be compressed, and a practice loop that matches how skill actually improves.
I write this for the compressed-timeline case. If you have six weeks, use a broader practice system. If you have fourteen days, the constraint is learning rate per hour, not content coverage.
Reading feels fast. Performance is not recognition.
Case interviews reward what you can produce under pressure, not what you recognize on a page. That distinction is old psychology, not consulting folklore.
In Roediger and Karpicke's 2006 testing-effect paper, students who repeatedly studied a prose passage scored well five minutes later. On a one-week delay, the group that had practiced retrieval by taking tests without restudy remembered about 61 percent of the material, versus about 40 percent for the study-only group. WashU's summary of the same study is here. Framework PDFs are the study-only condition. Spoken structure drills and timed math aloud are the retrieval condition.
So when someone asks for the "fastest" prep path, I translate the question: which activities force retrieval of the exact behaviors the interviewer is scoring? Anything that lets you nod along without producing an answer is slow, even if it fills a Sunday afternoon.
Deliberate practice, not case volume
Ericsson, Krampe, and Tesch-Römer (1993) defined deliberate practice as effortful, designed activity aimed at improving a specific aspect of performance, with feedback, not mere repetition of what you already do comfortably. Their violin and piano studies are about years of accumulated practice. The transferable claim for a two-week sprint is narrower: improvement tracks the quality of the feedback loop, not hours of unstructured play.
Mapped onto a case, that means three ingredients every session:
- A task that matches the interview behavior (speak a structure, compute aloud, read an exhibit, land a recommendation).
- A score or a tagged error you could defend to a partner.
- A redesign of the next rep so you are not rehearsing the same leak.
Running eight full mocks with no debrief is play. Running three mocks, each followed by a 15-minute fix on one scored dimension, is deliberate practice. Firms describe the same skill set without using the psychology jargon. McKinsey's interviewing page says the problem-solving interview evaluates how you structure ambiguity, work with facts and data, form conclusions, and articulate your thinking. BCG's case-interview preparation page lists structure, thoughtful questions, data analysis, quick calculations, and identifying the important factors. Those are the dimensions to score. Everything else is optional decoration when the calendar is short.
Ned's rule. If you cannot name the one scored behavior you will fix before the next rep starts, the last case did not count as practice. It counted as anxiety management.
A worked example: same 90 minutes, two learning rates
Suppose you have 90 minutes tonight. Candidate A watches a recorded mock and reads notes. Candidate B runs the loop below. Neither is fictional heroism; both schedules are reproducible.
| Block | Candidate A (passive) | Candidate B (scored reps) |
|---|---|---|
| 0–20 min | Watch a walkthrough | Structure drill: 4 prompts, spoken, 3 minutes each |
| 20–35 min | Skim a framework list | Tag one error (e.g. branches not MECE); rewrite one tree |
| 35–55 min | Light mental math on paper | Timed math aloud: 6 problems, formula first, then calc |
| 55–75 min | Read an exhibit solution | One exhibit: insight in 60 seconds, then numbers |
| 75–90 min | Feel prepared | One 8-minute mini synthesis; score recommendation clarity 1–5 |
Candidate B leaves with a written error tag and a second attempt that either fixed it or did not. Candidate A leaves with familiarity. On interview day, the interviewer is listening for Candidate B's behaviors. Familiarity does not appear on the sheet.
Hour math for two weeks: if you can protect 2 hours on weekdays and 4 hours on each weekend day, you get roughly 10 × 2 + 4 × 4 = 36 hours. That is enough for a Round-1-capable loop if almost none of it is passive. It is not enough to become a case specialist. Say that out loud before you start so you do not pretend otherwise.
The 14-day plan (about 36 hours)
Assume the interview is on morning of day 15. Protect the hours below. If your calendar only allows 24 hours total, cut volume before you cut the debrief: fewer full cases, same scored-fix loop.
| Days | Focus | Hours | What you produce |
|---|---|---|---|
| 1–2 | Case flow + opening script | 4 | Opening restated aloud 10 times; one clarifying-question habit |
| 3–5 | Structure + math isolation | 9 | 20 spoken trees; 40 timed math problems with formula-first |
| 6–7 | Exhibits + synthesis | 6 | 10 chart so-whats; 8 one-minute recommendations |
| 8–11 | Full spoken cases with scores | 12 | 6–8 full cases; after each, one tagged fix drilled once |
| 12–13 | Weakest dimension only | 4 | Re-drill the leak that still shows on your score sheet |
| 14 | Light rehearsal + rest | 1 | One short case or drill; sleep; no new frameworks |
Days 1–2 (4 hours). Learn the shape of a case: objective, clarifying questions, structure, analysis, recommendation. Do not collect ten frameworks. Practice the opening until you can restate the prompt, ask one constraint question, and announce your attack in under 90 seconds. A five-minute first rep on /start is enough to expose whether you freeze before you ever open a PDF.
Days 3–5 (9 hours). Isolate structure and math. For structure, profitability and market-entry prompts cover most first-round volume; invent MECE trees out loud, then write the one branch that was fuzzy. For math, state the formula before you touch a number. Use /drills for short scored reps so the feedback is immediate. Target: you can finish a clean tree in three minutes and a weighted-average or percent-change problem without going silent.
Days 6–7 (6 hours). Exhibits and synthesis. On every chart, force a so-what before you recite axes. On every synthesis, answer the client's question in the first sentence, then give one supporting number and one risk. If the recommendation arrives after a tour of your analysis, you are still narrating, not deciding.
Days 8–11 (12 hours). Full spoken cases. Prefer voice. The /interview path (CopperCup, about 18 minutes, seven-score debrief) exists for this phase: you need a score sheet, not a vibe. After each case, spend 15 minutes on one dimension only. Re-drill that dimension once before the next full case. Six well-debriefed cases beat twelve unreviewed ones on this timeline.
Days 12–13 (4 hours). Only the leak that still shows. If math setup is the problem, do not "do more cases." Do math. If structure collapses under pushback, do structure under a timer with a partner interrupting you.
Day 14 (1 hour). One light rep to keep the motor warm, then stop. Sleep is part of the plan. Cramming a new industry on the night before buys confidence theater and costs retrieval of the skills you already built.
What cannot be compressed
Be blunt about the limits, or the plan becomes a hustle brochure.
- Years of judgment. Deliberate-practice research is about accumulated, designed practice over long horizons. Two weeks can make your mechanics interviewable. It cannot manufacture ten years of business pattern recognition. Interviewers notice inventiveness; they also notice when a candidate has never seen a P&L. Do not fake the second.
- Behavioral depth. Personal-experience stories need real detail and rehearsal. You can tighten delivery in fourteen days. You cannot invent a leadership example you do not have.
- Calm under fire. Some anxiety drops with reps. Some does not. Plan for recovery language ("let me restart the tree from the objective") rather than for feeling zen.
- Firm-specific folklore. Public firm pages describe skills, not confidential scoring rubrics. Anything you heard on forums about "what BCG wants this year" is folklore until the firm says it. Treat it as optional color, not as a schedule item.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.
Your assignment for today
Do one 45-minute loop before you open another guide:
- Five minutes on /start to get a scored first taste without an account.
- Twenty minutes of structure or math on /drills.
- One spoken mini-recommendation from a short prompt you invent (company: HarborLine Shipping; question: cut cost 8 percent without cutting service).
- Write one error tag. Tomorrow's first ten minutes fix only that tag.
Everything on CoachNed is open for seven days with no card; then $120 for a recruiting season or $49 a month. Use the window for spoken reps, not for collecting more tabs.
Frequently asked questions
How long does it take to learn case interviews?
For a first-round-capable floor, Ned's estimate with a scored-rep loop is on the order of 30 to 50 focused hours for most strong candidates, often spread over several weeks. Two weeks at about 36 hours can get you interviewable mechanics if almost none of that time is passive reading. Becoming consistently strong usually takes longer; firms do not publish a required hour count.
Can I learn case interviews in two weeks?
You can learn to perform the scored behaviors: structure, math aloud, exhibit so-what, and a clear recommendation. You cannot compress years of business judgment or invent behavioral stories. Treat two weeks as a mechanics sprint with a hard stop-list, not as a promise of offer-ready polish.
Is reading case books faster than practicing?
Usually no. Retrieval practice outperforms restudy on delayed tests in the Roediger and Karpicke work cited above. Books help when they supply prompts you then speak. Reading solutions without producing answers is the slow path dressed as diligence.
Should I do full mocks every day on a short timeline?
Not until the pieces are legible. Early full mocks without a debrief rehearse panic. Once you can open, structure, and compute without freezing, full spoken cases with a scored fix between them are the highest-leverage use of remaining hours.
What should I skip if I only have two weeks?
Skip framework encyclopedias, rare industry deep-dives, and unreviewed video bingeing. Skip collecting competitor tips that are not on firm careers pages. Keep behavioral rehearsal on the calendar; cutting it to "save time for cases" is a common way to fail a round that was never only about the case.
Sources
- Test-Enhanced Learning (Roediger & Karpicke, Psychological Science, 2006) — retrieval practice vs restudy; delayed retention advantage. Checked 2026-09-24.
- WashU summary of Roediger & Karpicke testing-effect results — 61% vs 40% one-week recall figures from the same study. Checked 2026-09-24.
- The Role of Deliberate Practice in the Acquisition of Expert Performance (Ericsson, Krampe, Tesch-Römer, Psychological Review, 1993) — deliberate practice as designed, effortful improvement with feedback. Checked 2026-09-24.
- Interviewing at McKinsey — official description of problem-solving interview skills. Checked 2026-09-24.
- BCG Case Interview Preparation — official list of case skills (structure, questions, data, calculations, judgment). Checked 2026-09-24.