Classic consulting brain teasers—manhole covers, golf balls in a 747, three switches and a basement—have largely disappeared from McKinsey, BCG, and Bain interviews. Open any of those firms' own interviewing pages and you will find cases and personal-fit conversations, not riddle banks. What still shows up are questions that feel like teasers if you squint: market sizing, Fermi estimates inside a case, and short logic items on online aptitude tests. Those are scored as structured problem-solving, not as puzzle-camp cleverness. After this page you should know what to ignore, what to practice, and how an interviewer actually marks the surviving kind.
What the firms say they test
I do not have a confidential memo from any firm. I do have what they publish for candidates, and it is consistent.
McKinsey's interviewing page describes a personal experience interview and a problem-solving interview built around a typical client scenario. The sample cases on that site—Diconsa, Beautify, Electro-Light—are business problems with data and recommendations. There is no official "why are manhole covers round" module.
BCG's consulting interview process lists a skills interview, a case interview for client-facing roles, and team interviews. Its case preparation page says you will structure a realistic business challenge, ask questions, analyze data, do quick calculations, and synthesize. Again: case, not riddle.
Bain's interviewing overview says consulting candidates typically face a digital assessment and then case interviews. Bain's digital assessment hub puts the online step at roughly 30–40 minutes (Sova or TestGorilla, depending on office), aimed at problem-solving, communication, and analytical skill—not lateral-thinking parlor games.
If your invite matches those pages, spend your hours on cases, PEI stories, and the assessment format your recruiter named. A nostalgic partner can still ask a warm-up puzzle; that is candidate folklore, not the published process, and it is almost never the offer decision.
Google's public postmortem (useful, not consulting)
Consulting firms did not publish a tidy "we dropped brainteasers" press release the way tech did. The closest primary data point is Google. In a 2013 New York Times interview, Laszlo Bock, then Google's SVP of people operations, called classic brainteasers "a complete waste of time," said they "don't predict anything," and noted they mainly make the interviewer feel smart. Google shifted toward structured behavioral interviews with a consistent scoring rubric.
That is not a claim that MBB copied Google. It is evidence that large hiring machines that once romanticized puzzles later measured them and walked away. Strategy firms arrived at the same place for a different reason: the job is client problem-solving under time pressure, so the interview that looks like the job won.
What still looks like a teaser (and is not)
Three formats survive. Candidates mislabel all three as "brain teasers."
| Format | Where it shows up | What the sheet rewards | What it is not |
|---|---|---|---|
| Market sizing / Fermi estimate | Inside a case, or as a warm-up before exhibits | Structure, explicit assumptions, clean arithmetic, a sanity-checked range | A single clever trick answer |
| Quick estimation in a case | Mid-case ("roughly how big is that?") | Order-of-magnitude judgment while you talk | Memorized trivia |
| Short logic / numerical items | Online aptitude screens (e.g. Bain Sova/TestGorilla; other firms' Solve/CCA-style tests) | Accuracy under a clock, pattern recognition, composure | Open-ended riddling with the interviewer |
Market sizing is a standard case module. If someone calls "how many coffee shops in Austin?" a brain teaser, they are describing the surface of the question, not the scoring. The interviewer is listening for whether you segment, state rates, multiply without melting down, and land a number you would defend in a team room.
Aptitude logic is closer to timed GMAT-style reasoning than to "three light switches." You get a right answer or you do not. Talking through a beautiful wrong pattern does not help. That is why drilling riddles for an online test is almost always the wrong prep.
How I score the surviving kind
On a live market-size or estimate, I am not grading creativity theater. In a typical first-round case of eight candidates, two or three will freeze on the first multiplication; another two will invent a precise number with no segment tree. The ones who pass look boring on paper:
- Restate the question in one sentence.
- Name the segments or drivers before touching a calculator.
- State each assumption out loud and keep it round (10%, not 11.3%).
- Compute in chunks; announce intermediate totals.
- Sanity-check against a known anchor ("that would be $X per household; does that feel high?").
- Give a range or a base case plus the assumption that moves it most.
I stop listening when the candidate goes silent for ninety seconds, dumps a memorized manhole-cover speech, or treats the estimate as a trivia quiz about the "real" statistic. The real statistic is irrelevant in the room. The habit is the product.
Ned's rule. If the question has a cute one-line punchline, it is almost never how MBB hires in 2026. If it has a segment tree and a range, treat it as case math—because that is what it is.
Four worked examples (original numbers)
These are invented. Reproduce the arithmetic; do not memorize the answers.
1. Market size: weekday lunch bowls in a mid-size U.S. city
Prompt: "How many grain bowls does a lunch-focused fast-casual chain sell on a typical weekday in Portland, Oregon?"
Approach, spoken:
- Metro workers and students who buy lunch out on weekdays: start with ~1.0 million metro population; say 55% are working-age/students ≈ 550k; of those, 40% buy lunch out on a given weekday ≈ 220k meals out.
- Fast-casual share of lunch-out: maybe 25% ≈ 55k fast-casual lunches.
- Grain-bowl / salad-bowl niche of fast-casual lunch: say 15% ≈ 8,250 bowls citywide.
- One mid-size chain's share of that niche: say 8% ≈ 660 bowls/day.
Sanity check: 660 bowls × $14 average ≈ $9k daily revenue before drink attach. For a chain with a handful of urban locations, that is plausible; if I had assumed 5,000 bowls for one brand, I would know I overshot. Shakiest assumption: niche share. I say that last.
2. Estimation inside a case: warehouse overtime
Prompt mid-case: "Roughly how many overtime hours did the warehouse burn last quarter if five lines ran an average of 12 overtime hours each weekday?"
- Weekdays per quarter ≈ 13 weeks × 5 = 65.
- Hours: 5 lines × 12 hours × 65 days = 5 × 780 = 3,900 overtime hours.
That is not a riddle. It is whether you keep units straight while the interviewer watches. Round 13×5 to 65 without apology; do not invent a leap-year adjustment.
3. Fermi with a recommendation: ferry or more buses
Prompt: "Should Midcoast City add a peak ferry between the south pier and downtown, or put the same capital into bus lanes?"
This looks like a teaser. It is a tiny unstructured case. Tree:
- Demand: peak workers living south of the water; say 18,000; 20% would take a water option if it beat bus by 15+ minutes ≈ 3,600 potential riders.
- Capacity and cost: ferry every 20 minutes, 180 seats, three AM peak trips = 540 seats; fill at 70% ≈ 380 riders/day one-way. Capital: $40m vessel + dock vs $40m of bus-lane and signal work moving 2,000+ riders faster.
- Substitution: ferry steals from existing buses; net new transit trips may be far smaller than 380.
Land: unless the ferry cuts door-to-door time dramatically for a dense pocket the buses cannot serve, the bus package moves more people per dollar. The interviewer who wanted a riddle will cut you off; the one who wanted judgment will lean in.
4. Aptitude-style logic (not a live puzzle)
Prompt on a screen: "A warehouse ships 240 parcels in 8 hours with 6 packers. Same pace, how many packers to ship 450 parcels in 6 hours?"
- Rate: 240 / (8 × 6) = 5 parcels per packer-hour.
- Need: 450 / 6 = 75 parcels per hour → 75 / 5 = 15 packers.
On an online test, write 15 and move on. Do not narrate a witty story about packers. Speed and accuracy are the score.
How much teaser practice is rational
| Target list | Dedicated riddle hours | Better use of the same hours |
|---|---|---|
| MBB / Big Four strategy only | Zero, unless a recent office report says otherwise | Cases, PEI, named online assessment |
| Mixed list including econ / litigation consultancies | A few Fermi estimates per week, max | Still mostly cases; logic warm-ups second |
| Assessments only this month | None on classic riddles | Timed numerical and verbal practice for the named vendor |
If someone sends you a 44-puzzle PDF, treat it as entertainment. The transferable skill is the market-size habit above—and that habit is already case prep.
Practice this week (one assignment)
Do three original estimates out loud, ten minutes each, with a phone timer. Use new prompts: airport car-rental desks in a hub city; weekly meal-kit boxes for a 400k-household metro; hospital linen loads for a 300-bed hospital. For each, force a segment tree, two stated assumptions, and a sanity check. Then run a structured market-sizing drill so the habit sticks under mild pressure:
Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month. If you still need a full case after the estimates, start a five-minute first rep or sit the live voice case on /interview.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.
Frequently asked questions
Do McKinsey, BCG, or Bain still ask brain teasers?
Their published interviewing pages describe cases and fit (and, for Bain, a digital assessment). Classic riddles are not part of that advertised process. An individual interviewer can still ask a warm-up; treat it as composure practice, not as the core of your prep calendar.
Is a market-sizing question a brain teaser?
No. Market sizing is a standard case module scored on structure, assumptions, math, and a sanity-checked range. Calling it a brain teaser usually means the speaker is remembering Google-era folklore, not how consulting interviews are marked.
Should I memorize famous puzzles like the three switches?
No. Those answers leak, they do not look like the job, and they do not appear on the firms' own prep materials. If a logic puzzle appears on an online aptitude test, it will be a short, closed-form item—not a multi-minute live performance.
How is estimation scored differently from a riddle?
A riddle has a punchline. An estimate has a transparent method. Interviewers reward the method even when your final number is off by a factor of two; aptitude screens reward the correct closed answer under time.
What if my target firm is an economic consultancy, not MBB?
Logic and estimation warm-ups are more plausible there than at a standard MBB first round. Still confirm with your recruiter and recent candidates, and keep most of your hours on cases. Do not replace casing with a riddle syllabus.
Sources
- Interviewing at McKinsey — Official description of personal experience and problem-solving (case) interviews; sample client scenarios. Checked 2026-09-24.
- BCG consulting interview process — Skills interview, case interview, and team interview steps for consulting roles. Checked 2026-09-24.
- BCG case interview preparation — What BCG says a case tests: structure, questions, data, calculations, synthesis. Checked 2026-09-24.
- Bain interviewing — Consulting path: digital assessment plus case interviews. Checked 2026-09-24.
- Bain digital assessment — Sova or TestGorilla; ~30–40 minutes; problem-solving and analytical skills. Checked 2026-09-24.
- In Head-Hunting, Big Data May Not Be Such a Big Deal (NYT, 2013) — Laszlo Bock on Google dropping brainteasers as non-predictive. Checked 2026-09-24.