In a typical first round of eight, I hear "I'll structure this MECE" from five candidates. Two of those five then put the same dollar in two buckets. Another draws clean, non-overlapping buckets that still cannot isolate what moved the P&L. The interviewer is not scoring the acronym. The sheet rewards a partition that can find the driver.
MECE is not a framework you name. It is Barbara Minto's grouping rule for a pyramid of ideas: pieces that are mutually exclusive of each other and collectively exhaustive of the whole. After this page you should be able to fail each half on purpose, spot the "MECE but useless" trap, and recut a leaky tree under time.
Where MECE actually comes from
Barbara Minto developed the Pyramid Principle while editing McKinsey reports and later teaching structured writing across firm offices. In a McKinsey alumni interview, she says the grouping rule required ideas to be MECE — mutually exclusive and collectively exhaustive — and jokes that she invented the abbreviation (crediting Aristotle for the logic, herself for the label). She pronounces it as one syllable, "meece," rhyming with niece; most interviewers say "mee-see." Either is fine. Quoting the letters is not the skill.
Her own site describes the Minto Pyramid Principle as organizing thinking into a pyramid held together by a single governing thought, with lower-level facts grouped by similarity and summarized upward. The authoritative textbook is The Minto Pyramid Principle: Logic in Writing, Thinking and Problem Solving (1996, 12 chapters), which supersedes the shorter 1987 Pearson edition. McKinsey still lists that 1996 title among McKinsey on Books. Prep folklore that "McKinsey invented MECE last decade" is wrong; the public trail runs through Minto's mid-century firm work and her published pyramid logic.
What candidates get wrong: treating MECE as a brand of case structure. In Minto's terms it is the horizontal test on any layer of a tree — same kind of ideas, no overlap, full coverage of the parent. Name a dimension, then partition it. Do not announce MECE as if it were a product you bought.
The exclusive test (with a counterexample)
Mutually exclusive means one fact, one home. Ask: can the same unit of analysis (a customer, a dollar, a SKU, a process step) sit in two sibling buckets without stretching the labels?
If yes, you will double-count, ping-pong, or argue about taxonomy instead of the decision.
| Exclusive fail | Why it doubles | Recut |
|---|---|---|
| "Online" next to "retail customers" | Online is a channel; retail customers buy online | Split by channel or by customer type, not both at one layer |
| "Cost" next to "profitability" | Profit is revenue minus cost | Parent = profit; children = revenue drivers and cost drivers |
| "Short term" next to "quick wins" | Quick wins live inside short term | Nest quick wins under short term, or drop the synonym |
Counterexample. Candidate structure for "why did unit margin fall?": Product mix / Pricing / Competition / Costs. A price cut by a rival can live under Competition and under Pricing. A shift toward a low-margin SKU is Product mix and can also be reported as a Cost-per-unit story if the candidate muddles COGS. The exclusive test fails on the first pass. Recut on one dimension per layer — for example volume mix vs. price vs. cost per unit — and put competitor reaction under the branch that owns the mechanism you will test.
I stop listening when the candidate defends overlap as "they are related." Related is not exclusive. Nest or merge.
The exhaustive test (with a counterexample)
Collectively exhaustive means nothing material to this question is homeless. Ask: is there a fact that would change the recommendation with no bucket to put it in?
Exhaustive does not mean a long list. It means coverage relative to the decision and the data you are about to request. ESG, geopolitics, and brand can be holes only if they can move this answer. Otherwise they are scope theater.
| Exhaustive fail | What is homeless | Fix |
|---|---|---|
| Revenue only, on a margin case | Cost or mix that crushed margin | Add cost (or mix) as a sibling under profit |
| Passenger trips only, when cargo is 18% of revenue | Cargo P&L | Add cargo / other, or declare passenger-only and stick to it |
| "Existing customers" vs. "new logos" on churn | Mid-contract expansions that look like "new" ARR | Define the unit: logo, contract, or MRR, then partition that unit |
Counterexample. Prompt: Northpeak Fitness, a 40-club regional chain, asks why membership revenue is flat while the market grew. Candidate buckets: Marketing / Pricing / Competition. Exclusive might scrape by. Exhaustive fails: there is no home for retention (churn of existing members) or for capacity (waitlists and peak utilization that block new joins). Both would change the answer. Add a retention branch, or split revenue into new joins vs. renewals vs. ancillary, and you have a place for the missing facts.
Run both tests. Exclusive without exhaustive is tidy blindness. Exhaustive without exclusive is mush that covers everything twice.
The trap: MECE but useless
This is the mistake that looks polished and still scores poorly. Buckets can pass exclusive and exhaustive and still fail the interview because they do not isolate the driver.
Classic form: a candidate splits a $240m cost base into Fixed / Variable / Semi-variable, all exclusive, all exhaustive of cost. Then stares at three buckets of similar size and has no hypothesis. Or: Company / Customers / Competitors on a pure cost-out ask — MECE relative to "the business," useless relative to "where does the dollar leak."
The scoring heuristic I use: after the structure, can the candidate name one exhibit that would kill a branch? If every bucket needs "more research" equally, the tree is decorative. MECE is necessary for clean math and clean synthesis. It is not sufficient for insight.
Ned's rule. Before you defend a structure as MECE, ask whether any one bucket can absorb most of the variance if the data goes your way. If every bucket is equally likely, you built a taxonomy, not a case plan.
Useless MECE usually comes from cutting on a dimension that is easy to say (internal/external, short/long, people/process/tech) instead of the dimension the P&L or decision actually moves on (price vs. volume vs. cost; join vs. churn; lane A vs. lane B).
Worked example: repair a non-MECE tree
Prompt. Cedarline Pharmacy, a 28-store regional, saw prescription gross profit fall from $41.2m to $36.8m year over year (-$4.4m). The CEO wants the driver. Candidate opens with:
- Brand drugs
- Generics
- Specialty
- Insurance
- Walk-in retail front-of-store
That list fails exclusive immediately. Specialty is mostly brand or specialty-generic; insurance is a payer, not a molecule class; front-of-store is a different P&L line that may not even sit inside "prescription gross profit." The same script can land in Brand and Specialty. The same claim can bounce between Insurance and Generics ("payers forced generic substitution").
Step 1 — restate the parent. Parent = prescription gross profit (dispense margin), not total store profit. Front-of-store is out of scope unless the interviewer expands it.
Step 2 — pick one dimension for the first layer. For a margin drop, the clean first cut is usually volume (scripts) vs. margin per script, or mix of script types vs. rate within type. Mix both at the same layer and you recreate the mess.
Cedarline's first-layer repair:
| Branch | Definition | Exclusive? |
|---|---|---|
| Script volume | Number of prescriptions dispensed | Yes, vs. margin |
| Margin per script | Gross profit dollars / script | Yes, vs. volume |
Collectively exhaustive of prescription gross profit by identity: Profit = scripts x margin per script. No homeless third sibling called "insurance."
Step 3 — nest the next dimension under the branch that needs it. Under margin per script, split by molecule class using the store's own taxonomy (one label per script):
| Molecule class | Scripts (m) | Margin $/script | Gross profit ($m) |
|---|---|---|---|
| Standard brand | 0.42 | $28 | 11.8 |
| Standard generic | 1.95 | $8 | 15.6 |
| Specialty | 0.08 | $118 | 9.4 |
| Total | 2.45 | — | 36.8 |
Prior year for the same three classes: 2.38m scripts, gross profit $41.2m. Reproduce the bridge:
| Driver | Math | Contribution to -$4.4m |
|---|---|---|
| Volume | +0.07m scripts x prior blended margin ~$17.3 | roughly +$1.2m (helps) |
| Mix / rate | Residual after volume | roughly -$5.6m (hurts) |
Dig under margin: specialty scripts were flat; standard brand scripts fell 9% while generic rose. Brand margin per script also fell $3 on reimbursement cuts. The overlapping first structure could not isolate that. The repaired tree can: volume is not the problem; brand reimbursement and mix shift into generic are.
What to say in the room. "I will split prescription gross profit into script volume and margin per script, then cut margin by the store's brand / generic / specialty flags so one script cannot sit in two buckets. Front-of-store stays out unless you want total store profit."
That is MECE as Minto meant it: same kind of ideas, exclusive, exhaustive of the parent, ordered so a summary can sit above them.
When to stop polishing the buckets
Perfect partitions on the wrong question still fail. Restate the decision before you draw boxes. If the interviewer hands you exhibit categories that already partition cleanly, do not recut for sport. If you have thirty seconds and one obvious hole, a slightly coarse exclusive split beats a beautiful matrix you never use.
MECE is a quality test on a layer, not a reason to stall. The interviewer wants the next chart request, not a seminar on Aristotle.
Practice this today
Take any profitability or ops prompt. Draw your first-layer buckets in sixty seconds. Run the exclusive test (name one unit that could live in two places). Run the exhaustive test (name one fact that would change the answer with no home). Then ask Ned's rule: which single bucket could absorb most of the variance?
CoachNed's structure drills at /drills/structure score overlap and holes the way a first-round interviewer does. Everything is open for seven days, no card; then $120 for a recruiting season or $49 a month.
Frequently asked questions
What does MECE stand for in consulting?
Mutually exclusive, collectively exhaustive. Sibling buckets must not overlap, and together they must cover the parent idea for the decision at hand. Barbara Minto popularized the abbreviation as part of the Pyramid Principle; it is a grouping rule, not a named case framework.
How do you pronounce MECE?
Most consultants say "mee-see." Barbara Minto says she prefers one syllable, "meece," like niece. In an interview, either is fine. Clarity of the partition matters more than the phonetics.
Is MECE the same as the Pyramid Principle?
No. The Pyramid Principle is Minto's full method for organizing and communicating thinking (governing idea on top, grouped support below). MECE is the horizontal rule those groups must satisfy: exclusive pieces that exhaust the whole. You can recite MECE and still fail pyramid logic if the top point does not summarize the branches.
Can a structure be MECE and still be wrong for the case?
Yes. That is the MECE-but-useless trap: clean buckets that do not isolate the driver or map to a testable exhibit. Fix the dimension you cut on, not the spelling of the acronym.
Should I say "I'll be MECE" out loud in the interview?
Better to show it. State the dimension ("I will split margin into volume and rate, then rate by product class") so the interviewer can follow the partition. Announcing the letters without a clean cut is empty calories.
Sources
- Barbara Minto: "MECE: I invented it, so I get to say how to pronounce it" — Minto's account of coining MECE, pronunciation, and pyramid rules at McKinsey. Checked 2026-09-24.
- The Minto Pyramid Principle concept — Minto's own description of pyramid grouping and summarizing upward. Checked 2026-09-24.
- The Minto Pyramid Principle textbook (1996) — authoritative edition (12 chapters) vs. the shorter 1987/2002 Pearson book. Checked 2026-09-24.
- McKinsey on Books — firm listing of The Minto Pyramid Principle: Logic in Writing, Thinking, and Problem Solving (1996). Checked 2026-09-24.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.