Most candidates burn their first week of case prep on content that never shows up on a score sheet. They reread frameworks, highlight a case book, and feel prepared. I still score them as if they never opened the book. The interview grades what you do under time with incomplete data, not what you can recite after coffee.
That is the thesis of this piece. CoachNed's courses and learning-mode checkpoints teach the map of a case. Scored reps teach the room. Confuse the two and you walk into a first round with vocabulary and without a transferable structure. Finish this page knowing exactly what lives at /learn and /ai/learn, what a course cannot do, and which next rep closes the gap.
Ned built both. I am Edward "Ned" Calder. Partner in Big Four Strategy & Operations since 2018. The courses are the explanations I wish someone had given me before I froze in my own first case.
What ships at /learn today
Open Courses and you get seven written lessons, not a video library and not a peer marketplace. The page states it plainly: finish the first two, then sit a drill. Lessons 03–07 sit on paid plans after the free week ends.
| Step | Lesson | Time | Access (after free week) |
|---|---|---|---|
| 01 | The Case Interview Method | ~8 min | Free forever |
| 02 | Market Sizing | ~8 min | Free forever |
| 03 | Case Frameworks | ~9 min | Season Pass / Pro |
| 04 | Profitability Cases | ~8 min | Season Pass / Pro |
| 05 | Case Math Essentials | ~9 min | Season Pass / Pro |
| 06 | Structured Brainstorming | ~6 min | Season Pass / Pro |
| 07 | Behavioral & McKinsey PEI | ~8 min | Season Pass / Pro |
Lessons 01 and 02 together run about sixteen minutes. The method lesson is the 60-second loop: clarify, structure, hypothesize, math, close. Market sizing covers top-down versus bottom-up with worked estimates. That is the free on-ramp. Everything else on the list is short, written, and meant to be used once before you leave the page and do a graded task.
Separately, How AI actually works is a free AI-literacy lesson of about twelve minutes. It is not one of the seven casing lessons. ML, LLMs, RAG versus fine-tuning, agents and harnesses — in English — then SurePath Mutual while the language is still warm. You do not need to train a model. You need to talk about one without bluffing.
Learning mode is not the course catalog
The courses are the written map. Learning mode is a case-player setting on selected cases: short checkpoints before the live turns. On BrightTrail, that means six checkpoints that force a commit — clarify, structure, brainstorm, exhibit, synthesis — before Ned throws you into the full prompt. Juniper Cup runs the same idea on a growth case. The how-to-case page and the method lesson both point you there on purpose.
| Surface | Job | Output |
|---|---|---|
| /learn courses | Teach vocabulary and the decision loop in writing | You can narrate the stages without notes |
| Learning mode on a case | Walk checkpoints with teaching text, then sit the case | You have seen a strong move at each gate |
| Guided / voice case | Produce answers under time and take a seven-score debrief | A number on structure, math, exhibits, synthesis |
| Skill drills | Isolate one leak for five to ten minutes | A graded micro-rep on that dimension |
A course that ends without a scored output is a blog post with a login wall. Learning mode that never graduates into guided or voice is a crutch. I built the sequence so the map and the checkpoints are short on purpose. The product that moves the offer rate is the debriefed rep.
What deliberate practice actually requires
Ericsson, Krampe, and Tesch-Römer’s 1993 paper in Psychological Review defined deliberate practice as effortful, feedback-rich work designed to improve a specific performance, not as hours of comfortable repetition. Their Berlin violin and piano studies tied higher performance to accumulated deliberate practice under those constraints, not to talent slogans. You can read the framework in their Psychological Review article (PDF).
Two implications for casing:
- A well-designed task with feedback beats volume of passive study. Rereading the profitability tree for the third time is not deliberate practice. Writing a tree for HarborSnack, getting scored, and rewriting the weak branch is.
- The feedback has to name the next action. “Good structure” is useless. “Your volume branch mixed tickets with stores; rebuild volume as tickets × stores × hours” is practice.
Roediger and Karpicke’s 2006 Psychological Science experiments on test-enhanced learning make the same point from the memory side. In their delayed-retention condition, students who repeatedly retrieved a prose passage remembered more a week later than students who restudied it, even though restudy felt better in the moment. Their paper is freely available as a PDF. Reading frameworks raises confidence. Retrieving a structure under a blank prompt raises performance.
The National Academies’ How People Learn II (2018) synthesizes the broader literature: encoding, consolidation, and retrieval are different processes, and fluent retrieval is the goal for performance settings. A case interview is a retrieval setting with a partner watching the clock.
McKinsey’s own careers page describes the problem-solving interview as a discussion of a client scenario used to see how you structure ambiguity, handle facts, and recommend. That is performance, not a quiz on frameworks. See Interviewing at McKinsey. CoachNed is independent of McKinsey and every other firm named here; I am describing the public format, not a confidential score sheet.
What a course cannot do
Be blunt about the limits. I wrote the lessons. I still will not let you treat them as enough.
A course cannot put you under interviewer pressure, grade spoken delivery, fix a math leak without timed equations, replace PEI follow-ups, or guarantee an offer. Sixteen minutes of reading is not a twenty-five-minute interrupted case. Lesson 05 lists formula families; Quick math forces retrieval. Lesson 07 sketches STAR; the behavioral simulator asks the second “what did you do?” Firms do not publish cut scores for good reason.
If you finish all seven lessons and skip scored cases, you have learned about casing the way someone learns swimming from a pool-safety poster. Useful. Incomplete.
Worked example: HarborSnack in twenty minutes
Invented client, reproducible numbers. HarborSnack sells branded trail mix. Revenue last year was $96m. Contribution margin was 28%. Fixed costs were $18m. Operating profit was therefore:
Profit = (0.28 × 96) − 18 = 26.88 − 18 = $8.88m.
This year they ran a volume promo. Revenue rose 8% to about $103.7m. Contribution margin fell to 24% because of trade spend. Fixed costs stayed $18m.
New profit = (0.24 × 103.7) − 18 ≈ 24.9 − 18 = $6.9m.
Top line up. Profit down about $2m. The recommendation is not “grow volume.” It is protect price and kill the next deep-cut promo until you see unit economics by SKU.
| Prep choice (Ned’s estimate) | What you practice | What I would score on Monday |
|---|---|---|
| Reread profitability frameworks for 40 minutes | Vocabulary | Still freezes on the bridge |
| Read lesson 04 once (~8 min), then rebuild HarborSnack’s bridge yourself | Retrieval of the profit equation | Passable first pass |
| Learning-mode checkpoints on BrightTrail, then one guided rep | Commit → teach → independent answer | Usable structure plus a debrief leak |
| Two scored math drills on contribution and mix | Timed retrieval | The promo mistake dies before the interview |
The map told you to disaggregate revenue and cost before brainstorming. The scored rep is where you notice you still jumped to “marketing ideas” before the $2m hole.
Ned's rule. Spend at most one timed block on a lesson for every three timed blocks on scored output. If the ratio flips, you are collecting confidence, not skill.
How I would sequence a first week
Assume you have never run a full case end to end.
- Lesson 01 at /learn/the-case-interview-method (about eight minutes). Close the tab.
- BrightTrail in learning mode: hit the checkpoints, then finish the case turns and take the debrief.
- One structure drill on whatever the debrief named.
- Lesson 02 on market sizing only if sizing showed up as a blank. Otherwise skip it until you need it.
- If your first round includes an AI-flavored ops or claims case, read /ai/learn once and sit SurePath Mutual the same day.
- From day three onward: guided or voice cases, with drills on the weakest dimension. Lessons 03–07 when a gap appears, not as a binge.
Access, as the pricing page states it: everything open for seven days, no card; then $120 for a recruiting season (Season Pass, 90 days) or $49 a month (Pro). After the week, Quick math, the sample walkthroughs, the AI lesson, and the first two courses stay free. Do not buy a plan to finish the curriculum. Buy it to finish forty scored cases before your first round.
When to leave the map
Graduate from courses and learning-mode checkpoints when most of this list is true without notes:
- You can restate the decision and ask two clarifying questions that change the tree.
- You can draw a two-level MECE tree in roughly 90 seconds and name the first branch to test.
- You write the equation before the numbers and attach a so-what.
- You read an exhibit by stating the implication, not by narrating the axes.
- You close with recommendation first, two supports, one number, one risk, one Monday step.
If you can do that on paper but still ramble out loud, move to voice. If one dimension keeps failing, stay in drills. Do not open lesson 03 because it feels unfinished. Open it when a debrief says your tree is generic.
Do this today
Block forty minutes.
- Read The Case Interview Method once. Do not take notes past three bullets.
- Open BrightTrail in learning mode from /consulting-cases, finish the checkpoints, then the case.
- Run one structure or math drill named by the debrief — or start with Quick math if the leak is speed.
If you want the five-minute no-account taste first, use /start. If you want Ned asking out loud, use /interview. Pricing detail lives on /pricing.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.
Frequently asked questions
What is CoachNed Learning Mode?
Learning mode is a case practice setting with short teaching checkpoints before the full turns. On BrightTrail that is six checkpoints. It is separate from the seven written courses at /learn, though the method lesson points you into learning mode on purpose.
Are the courses free?
Lessons 01 and 02 (method and market sizing) stay free after the trial week. Lessons 03–07 require a Season Pass or Pro once the free week ends. The AI literacy lesson at /ai/learn stays free. During the free week, everything on the product is open.
How is a course different from a scored case?
A course explains the move in writing. A scored case forces you to produce the move under time and returns seven dimension scores plus a named next drill. Interviewers hire the second skill.
Should I finish all seven lessons before my first case?
No. Finish lesson 01, sit a learning-mode case, then use later lessons only when a debrief names that gap. Bingeing 03–07 without reps is the confidence trap Roediger and Karpicke documented for restudy.
What is the AI lesson for?
/ai/learn is the generalist bar for AI language in cases: ML versus LLM, RAG versus fine-tune, agents versus harnesses, error cost instead of demo accuracy. Read it once, then sit SurePath Mutual. Specialist technical loops go deeper than this module.
Sources
- The Role of Deliberate Practice in the Acquisition of Expert Performance (Ericsson, Krampe, Tesch-Römer, 1993 PDF) — definition of deliberate practice and the Berlin musician studies. Checked 2026-09-24.
- Test-Enhanced Learning (Roediger & Karpicke, 2006 PDF) — retrieval practice outperforms restudy on delayed retention. Checked 2026-09-24.
- How People Learn II: Learners, Contexts, and Cultures (National Academies, 2018) — synthesis of encoding, consolidation, and retrieval for performance settings. Checked 2026-09-24.
- Interviewing at McKinsey — public description of the problem-solving (case) interview as a client-scenario discussion. Checked 2026-09-24.