Pyramid Principle

Lead with the answer. Group the supports. Make the supports MECE. Order the supports by deductive or inductive logic. Repeat one level down. The deck is a pyramid: one governing thought on top, three to five MECE supports under it, three to five MECE sub-supports under each — and that's the structure of every consulting document written since 1973, whether the team knows Minto or not.

Where it comes from

In 1963, Barbara Minto joined McKinsey as the firm's first female professional. She was sent to the London office to teach the European consultants how to write — McKinsey's American consultants were considered the firm's gold standard, and the partners wanted that quality replicated overseas.

What Minto found, in document after document, wasn't a writing problem. It was a thinking problem. The consultants were stacking observations chronologically — this is what we did, this is what we found, this is what we conclude — and the conclusion never landed because the reader hadn't been told what question the document was answering until page 27. The structure of the page was the structure of the work, not the structure of the reader's brain.

So Minto inverted it. Put the answer first. Then the few — three to five — reasons that justify the answer. Then, under each reason, the few facts that justify it. The shape of the document is a pyramid: one governing idea on top, and supporting ideas branching beneath in groups that are mutually exclusive (no overlap) and collectively exhaustive (nothing missing). She codified this in internal training materials at McKinsey through the '60s and early '70s, then published the canonical version as The Pyramid Principle: Logic in Writing and Thinking in 1973.

The book, expanded several times since, became the implicit operating manual of every top-tier strategy firm. Walk into a McKinsey, BCG, or Bain meeting today and the deck on the table will be a Minto pyramid whether anyone in the room remembers her name.

What it actually is

The pyramid is built on three rules — Minto stated them precisely in the book; they are the spine of the method.

1. Ideas at any level must be a summary of the ideas grouped beneath them. The top of the pyramid is one sentence. The three to five things under it, taken together, prove that sentence. The three to five things under each of those, taken together, prove it. There is no slide in the deck that doesn't ladder up.
2. Ideas in each grouping must be the same kind of idea. Don't list "three reasons" where two are causes and one is a consequence — that's logical asymmetry, and the reader can feel it even if they can't name it. Each group is a clean category.
3. Ideas in each grouping must be logically ordered. Either deductive (premise → premise → therefore conclusion) or inductive (three pieces of evidence sharing a common cause). One structure per group. Mixing them inside a single grouping is what makes a document feel "muddled".

Beneath those three rules sit two canonical patterns from the same book that get cited as standalone tools but are really sub-elements of the pyramid:

MECE — Mutually Exclusive, Collectively Exhaustive. The discipline that makes any single grouping work. If your three revenue-growth levers overlap (e.g. "raise prices" and "improve pricing strategy"), the pyramid leaks. If they don't cover the universe (you list "raise prices" and "cut costs" but forget "win new customers"), the pyramid lies.

SCQA — Situation, Complication, Question, Answer. The four-beat opening Minto designed for the document's introduction. The Situation anchors the reader in something they already accept; the Complication introduces tension; the Question is the one the reader is now asking; the Answer is the top of your pyramid. SCQA is what you say before the pyramid; the pyramid is what you say after.

Why it works

It works because it solves the two failure modes of expert communication at once.

The first failure mode is the long climb. Experts naturally tell stories in the order they discovered things — first the data, then the analysis, then the implication. Readers don't want that. Readers want the answer, then enough scaffolding to trust the answer, then out. A pyramid lets a busy reader stop at any level — read the first page (the governing thought), or read the first plus the supports, or read everything — and get the right amount of information for their need. The deck is a progressive disclosure, not a sequential revelation.

The second failure mode is logical mush. When supports overlap, or when a "reason" is actually a cause and another is a consequence, the reader's brain works overtime trying to flatten the asymmetry into a list. They feel the deck is "confusing" without being able to say why. MECE groupings remove that load. The reader doesn't have to hold the structure in working memory — the structure holds itself.

Together, these are the cognitive economics that make the pyramid durable: minimum reader effort per unit of conviction transferred.

When it's the right tool

The pyramid fits when the document has a single governing decision or recommendation the reader is being asked to accept, and the case for it can be decomposed into a small number of clean groups. That covers most strategy decks, most board memos, most investment committee papers, most consulting deliverables.

It is the wrong tool for genuinely exploratory work — early discovery decks, problem-definition workshops, hypothesis-tree sessions — where the answer hasn't crystallised yet and trying to write a top-of-pyramid is dishonest. It is also the wrong tool for narrative-driven communication where surprise is part of the value (keynote speeches, fundraising pitches built on reveal). The pyramid ends a story before it begins; sometimes that's not what you want.

A useful diagnostic: if you can write the deck's title as a single declarative sentence with a verb in it, you are ready for the pyramid. If you can't, the work isn't done yet.

Worked example — the cost-cut recommendation

Imagine the CFO is recommending a 15% cost cut to the board. The analysts have spent six weeks on the analysis. The temptation is to walk through the analysis chronologically.

The wrong shape (chronological):

Section 1 — Methodology. We benchmarked 14 peers across four SG&A functions… (40 slides later) Section 4 — Recommendation. Cut SG&A by 15% over 24 months.

By section 4, the audience has lost the question.

The right shape (Minto):

• Governing thought (slide 1): Cut SG&A by 15% over 24 months to recover 200 bps of operating margin.
• Three supports (slides 2–4, MECE — three buckets that together cover the savings):
  1. Procurement renegotiation — 60 bps
  2. Span-of-control compression — 80 bps
  3. Vendor consolidation — 60 bps
• Under each support, three sub-supports (slides 5–13) — the evidence that each lever delivers what's claimed.

The reader sees the answer on slide 1. They scan slides 2–4 to satisfy themselves that the three buckets are MECE and add up. They dive into 5–13 only on the bucket they're sceptical about. The 30-minute board conversation lands; the 60-slide chronological version would not.

The introduction to that pyramid — what Minto called the SCQA opening — would read:

Situation: Operating margin has declined 300 bps over five years. Complication: Pricing power is constrained; growth alone won't close the gap. Question: Where does the margin come back from? Answer: A 15% SG&A cut over 24 months recovers 200 bps.

That's the bridge between the reader's mental model and the top of the pyramid. Two sentences land you at the answer.

Common pitfalls

The fake pyramid. A title slide that says "the answer" but isn't supported by what follows. Three supports that don't actually defend the headline. The reader reads the headline, looks at the supports, and the headline doesn't survive. This is worse than no pyramid — it broadcasts confidence the analysis can't back.

Non-MECE supports. Three reasons where two overlap or three that don't cover the universe. The most common failure mode in real decks. A useful test: can you state what each support does NOT include? If you can't, it's not exclusive.

Tacit logic. The pyramid hides the deductive vs inductive structure. The reader can't tell whether your three supports are three pieces of evidence for one cause (inductive) or three steps of an argument (deductive). State it. "For three reasons:" signals inductive. "Therefore:" signals deductive.

Chronology pretending to be a pyramid. "Phase 1, Phase 2, Phase 3" is not a pyramid — it's a timeline. The pyramid is about logical hierarchy, not temporal hierarchy. A timeline can sit inside the pyramid (under one support), but it can't be the pyramid.

Top-of-pyramid that isn't an answer. "Strategic options for growth" is a topic, not an answer. "Pursue Option B (regional acquisition) over A (organic) because it captures 80% of the upside in half the time" is an answer. The discipline is to keep rewriting the headline until it has a verb and a defensible claim.

Related canon

• governing-thought — the one-sentence top of the pyramid. The pyramid is the structure; the governing thought is the apex. In progress.
• scqa-framework — the four-beat opening Minto designed to bridge from reader context to the pyramid's apex.
• action-titles — the slide-level application of the pyramid: every slide's title is a sub-claim that, taken together, proves the deck's title.
• big-idea-formula — Nancy Duarte's narrative cousin to the governing thought.

Canonical references

• Barbara Minto — The Pyramid Principle: Logic in Writing and Thinking — Pearson, 1973 (revised 2009).
• Barbara Minto — The Minto Pyramid Principle (author site) — primary source.
• Ethan Rasiel — The McKinsey Way — McGraw-Hill, 1999 — operationalises Minto for in-firm use.
• Gene Zelazny — Say It with Charts — McGraw-Hill, 1985 — the visual companion to Minto's textual pyramid.
• Tom Wujec — The Pyramid Principle — HBR, 2015 — modern restatement.
• McKinsey — The structured problem-solving and communication approach — internal-style overview.