Bain's official mock case interview runs 28 minutes, and by my count the interviewer speaks with a purpose eighteen times. Those eighteen moments are the most useful thing in any case interview video, and almost nobody watches them. A case video is a recording of two people, and candidates study the wrong one.
The candidate on screen is one data point: one person, one case, one day. The interviewer is the scoring sheet being filled in aloud. Every "why those three?", every "what will this tell us?", every exhibit handed over instead of the data you asked for is a line on the sheet moving.
Below: a log of every intervention in Bain's mock, a decoder from intervention to sheet line, and a worked example you can score yourself. After that, any mock, and any recording of you, reads like a graded transcript.
What the firms actually put on camera
Prep sites talk about "MBB case videos" as if the firms published a library. They do not. This is what sat on each firm's own pages on 2026-09-24.
| Firm | On its own pages | Full mock on video? |
|---|---|---|
| Bain | Mock Case Interview at Bain on its YouTube channel, 28 minutes, December 2023; written cases Coffee Shop Co. and FashionCo on its case interview preparation page | Yes |
| BCG | A case interview preparation page: what to expect, a do-and-don't list, the free OneDay@BCG job simulation, a short tips video from an Associate | No |
| McKinsey | An interviewing page with written practice cases (Diconsa, Electro-Light) containing "logic, thought processes, and suggested answers," plus a recruiting webcast | No |
| Deloitte | A case and scenario interview tips page: a five-step approach and a link to an interactive prep tool that calls itself a 15-to-20-minute exercise, not an assessment | No |
| Oliver Wyman | Written practice cases with difficulty ratings and sample answers (Dairy Farm, Autism Digital Device, Oil Gas Price Strategy) | No |
So "official case interview video examples" is, in practice, one video, with roughly 560,000 views when I opened it. The firms do not pretend their materials are complete: Bain's prep page points candidates to four commercial prep vendors, and BCG's says "there isn't always a single 'right' answer—what matters most is how you approach the problem and the quality of your reasoning." Pages change; open the links before you rely on this table.
Eighteen interventions in Bain's own mock
The case: FoodCo, a multi-billion-dollar food company, is weighing the acquisition of Smilk, an alternative-milk brand with about 6% US share, and wants to know whether Smilk's profit can grow sixfold in five years. The description says the case interview "is used at every round of Consultant interviews globally"; the interviewer says the case is based on a real engagement. It is a produced demonstration, not a live interview. I logged every time the interviewer spoke with a purpose, "great"s and "mhm"s excluded. Your count may differ by one or two; the disagreement is the exercise.
| # | What the interviewer did | What it scores |
|---|---|---|
| 1 | Says they will "be the one pushing us along" and own the time | Frame. Who drives the agenda |
| 2 | Candidate asks about gaps in the target's range: "not that it's relevant to the case currently" | Deflection. Clarifying questions score on relevance, not count |
| 3 | Candidate asks about the acquisition thesis; the answer adds that share matters as much as profit, and "the trade-offs between the two" | Gift. A relevant question is paid in data; the hint returns at #12 |
| 4 | "Before we jump into your framework, can you tell me why you thought about these three?" | Why-probe. Reasoned structure, or recited |
| 5 | Asks what the most important elements inside the target would be | Depth probe. Prioritization before a bucket is opened |
| 6 | "I like how you're bringing it back to the objective around this 6X" | Named reward. A named behavior is on the sheet |
| 7 | Hands over a market chart and asks what it says about the dynamics | Exhibit. Trend first, then the number that matters |
| 8 | Candidate asks for internal financials; gets a competitor share view instead | Redirect. The interviewer's agenda beats the request |
| 9 | "What else would you need to know about Smilk to gain confidence that they could gain share?" | What-else. The first list was judged incomplete |
| 10 | Third exhibit, a table with blanks: what do you see, and what would you need | Two-part exhibit. Reading plus gap-spotting |
| 11 | Before any arithmetic: "what is it going to tell you about whether or not they can achieve that 6X?" | So-what first. Knowing what the math will prove |
| 12 | "How are you thinking about the trade-offs, though, some of the headwinds they might face?" | Challenge. Share versus margin, planted at #3 |
| 13 | "So glad you're acknowledging that; why don't we go ahead and do the exercise" | Release. Permission to compute, after the so-what and the trade-off |
| 14 | After the math: "we are asking for a lot to go right, share gain and margin"; fourth exhibit | Tempering. Over-claiming from a sensitivity table |
| 15 | Candidate asks for willingness-to-pay data; the team "did do" that, now a portfolio question instead | Pivot. Creativity on demand; the data request is parked |
| 16 | Candidate proposes a roll-up; the fast-growing rivals' margins may "look very negative" | Counter-fact. Absorb it rather than defend the idea |
| 17 | "Knowing we don't have all the information that we typically would have, where is your recommendation?" | Close call. Committing under incomplete information |
| 18 | Reveals the real client found no willingness to pay a higher price | Reveal. Unscored; the right next step beat the right answer |
Folklore files Bain under "candidate-led." In Bain's own mock the interviewer announces in the first minute that they will manage the time, hands over four exhibits in their own order, and twice declines the data the candidate asks for (#8, #15). The candidate drives the reasoning; the interviewer drives the agenda. Expect both, anywhere.
The decoder: what each intervention means for you
Every firm words its sheet differently and none publish it, so the lines below are generic. The moves are not.
| When you hear | Sheet line | What to do |
|---|---|---|
| "Why those buckets?" | Structure | Give the one-line logic per bucket before anyone asks |
| "What will this tell us?" before math | Hypothesis-driven analysis | Say the threshold: "if market growth alone gives 2X, the rest must come from share or margin" |
| "What else?" | Breadth | Have a second list ready, organized by category |
| An exhibit you did not ask for | Drive versus control | Stop pulling and read; in a first round this is usually the clock, not a penalty |
| "How are you thinking about the trade-offs?" | Judgment | Check your notes; it was probably planted earlier |
| "Where is your recommendation?" plus a caveat | Synthesis | Answer, two reasons, one risk, one next step |
| Silence | Everything | The best sign there is |
That last row matters most. The Bain interviewer says "that's right" and "correct" after the chart reads because the video is teaching. In a real first round I rarely confirm anything; I speak when a candidate has stalled, drifted, or over-claimed. If I am quiet while you work, you are scoring.
Ned's rule. Count the interviewer's unprompted sentences, not their compliments. Every time an interviewer has to speak before you have finished a thought, a line on the sheet has moved, and it rarely moved up.
One more cut helps: is the intervention corrective or forward? A corrective probe pulls you back to something you should have done ("why those three?"). A forward probe extends a thought you had ("where else could they take the portfolio?"). By my reading, four of the Bain probes are corrective and the rest forward. That ratio, more than the compliments, is what a pass looks like.
A worked example you can score yourself
Larkspur Dental Partners is invented; the figures are round on purpose. Forty clinics, revenue of $96 million and flat for two years, EBITDA margin down from 18% to 12%. The client wants to know why, and whether 18% is recoverable within a year. Both candidates get the same exhibit.
| Cost line, share of revenue | Two years ago | Now |
|---|---|---|
| Clinical and front-desk staff | 50% | 56% |
| Supplies and lab fees | 12% | 13% |
| Rent, equipment, other | 20% | 19% |
| EBITDA | 18% | 12% |
Candidate A starts computing. "Eighteen percent of 96 is 17.28, twelve percent is 11.52, so they've lost about 5.8 million." Interviewer: "Before you go on, what is this going to tell you?" A: "How much they lost." Interviewer: "Which line would you look at?" A: "Staff, it went up." Two corrective probes in forty seconds, and the arithmetic was perfect.
Candidate B reads first. "Revenue is flat, so this is a cost problem, and it is almost entirely staff: six points, the whole margin loss. Supplies and rent cancel. Six points of $96 million is $5.76 million. If staff is the driver, the fix is wages or productivity, not marketing, so I want staff cost per clinic: 56% of 96 is 53.76, over 40 clinics is $1.34 million, against $1.2 million before, up 12%. To get back to 18% with costs held flat, revenue must rise until $84.48 million of cost is 82% of it: 84.48 over 0.82 is about $103 million, roughly 7% more. A 7% price rise with no volume loss is aggressive for dental, so I would look at productivity first." The interviewer says nothing until B stops, then: "What would you want to know before recommending a price increase?" One forward probe.
Check the numbers: 96 × 0.06 = 5.76; 96 × 0.56 ÷ 40 = 1.344; 84.48 ÷ 0.82 = 103.0; 103 ÷ 96 − 1 = 7.3%.
| Sheet line (0–3) | Candidate A | Candidate B |
|---|---|---|
| Structure of the read | 1 | 3 |
| Math accuracy | 3 | 3 |
| Narration: so-what before compute | 0 | 3 |
| Link to the objective (18% in a year) | 1 | 3 |
| Drive: who proposed the next step | 1 | 2 |
| Total | 6 of 15 | 14 of 15 |
A's math scored full marks and A finished at 6. B dropped a point because the forward probe went after something B had not asked: whether flat revenue hid falling patient volume behind higher fees. You could have scored both from the interventions alone, two corrective against one forward. That is the skill the videos are for.
Why a video feels like practice and is not
Watching a mock feels productive because it is fluent: the candidate never stalls the way you do, the numbers round, the exhibit arrives on cue. That fluency is the problem. In a well-known Psychological Science study, Roediger and Karpicke had students either re-read a passage or take recall tests on it. A week later the re-readers remembered 40% and the tested group 61%, although the re-readers had seen the passage 14.2 times against 3.4, and the re-readers were the more confident group.
Watching a case video is re-reading. Being asked "what else?" by someone who then waits is the test. A season of videos produces what the study predicts: high confidence, thin recall, a first round that feels unfair.
My estimate, from candidates I have debriefed: three to five videos in a season, each with the log, is plenty. After that, one video per three live cases at most, to recalibrate the decoder, never to learn content.
Run the log on yourself today
One assignment, three parts, about an hour.
- Open the Bain mock with a sheet of paper. Log every intervention with a letter (F frame, D deflection, W why-probe, X what-else, S so-what-first, R redirect, C close). Compare with my eighteen. Where we disagree, decide who is right; that argument teaches more than the video.
- Sit one live case on /interview: UrbanBrew, about 18 minutes, seven-score debrief. Before reading the debrief, write down every time the interviewer had to speak before you had finished a thought, corrective or forward, then see how well your log predicted the scores. Short on time? /start is three typed turns with instant scores and needs no account.
- Drill whatever the log shows most. Why-probes point to /drills/structure; "what will this tell you?" points to /drills/charts and /drills/math; a weak close points to /drills/synthesis.
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 Bain, BCG, McKinsey, Deloitte, or Oliver Wyman; the log above is my reading of a public video, not knowledge of any firm's sheet.
Frequently asked questions
Are there official case interview videos from McKinsey, BCG, or Bain?
As of 2026-09-24, Bain is the only one of the three with a full-length mock on its own channel, the 28-minute video logged above. BCG's careers page carries a short tips video from an Associate; McKinsey's interviewing page offers written practice cases with suggested answers and a recruiting webcast. All three pages change; open them rather than trusting a summary, including this one.
Are the mock case interviews on YouTube realistic?
Firm-produced mocks show the format faithfully and the interviewer's moves are real, but the confirmations are more generous than in a live round because the video is teaching. Peer-recorded mocks vary, and the interviewer is often a classmate who does not know what the sheet rewards. Log them anyway: an interviewer who never intervenes is telling you the video contains no information.
Can I learn case math from watching videos?
You can learn the narration order, which matters: in the Bain mock the candidate asks permission before rounding the market size, announces when 3.8 billion becomes 4 billion, and the interviewer lets both pass. You cannot learn speed by watching. That is retrieval, and it comes from timed reps such as /drills/quick-math, which is free for as long as you want it.
Sources
- Mock Case Interview at Bain — Bain & Company's official 28-minute mock, December 2023; the intervention log, the case facts, and the every-round statement. Checked 2026-09-24.
- Case interview preparation | Bain & Company — the Coffee Shop Co. and FashionCo cases and the four external vendors listed. Checked 2026-09-24.
- Case Interview Preparation | BCG Careers — what to expect, OneDay@BCG, the Associate video, the quoted sentence. Checked 2026-09-24.
- Interviewing at McKinsey — written practice cases with suggested answers and the recruiting webcast. Checked 2026-09-24.
- Case and scenario interview tips | Deloitte US — the five-step approach and the link to the interactive prep tool. Checked 2026-09-24.
- Interview preparation | Oliver Wyman — practice cases with difficulty ratings and sample answers. Checked 2026-09-24.
- Test-Enhanced Learning, Roediger and Karpicke, Psychological Science, 2006 (PubMed) — study design; study raised confidence, testing raised retention. Checked 2026-09-24.
- Test-Enhanced Learning, full text (Purdue Learning Lab PDF) — the 61% versus 40% recall figures and the 3.4 versus 14.2 readings. Checked 2026-09-24.