The Minto pyramid: start with the answer, with a worked example

The pyramid principle is a communication method developed by Barbara Minto, a former McKinsey consultant, and set out in her book "The Pyramid Principle". Its central idea is that a reader or listener understands better when they receive the main idea before the details, because they then know what to attach each following piece of information to.

The method rests on three rules.

• The answer first. The top of the pyramid is one sentence that answers the question asked.
• Each level summarises the one below. An argument is the synthesis of the facts filed under it, not just a heading. "Costs" is not an argument; "closing these stores costs less than keeping them" is.
• Ideas on the same level form a logical set. They are of the same kind, do not overlap and cover the subject, which amounts to saying they are MECE, a term we also owe to Barbara Minto.

In consulting, this is called "top-down" communication. It is the expected form for the closing synthesis of a case, and more broadly for every answer you give during the interview.

The pyramid reads from top to bottom when you speak, and is built from bottom to top when you think. You start from the facts the case has produced, group them by what they demonstrate, write one sentence per group, then look for the single sentence that these arguments support together.

A simple test checks each link. From top to bottom, each level should answer the "why?" or "how?" raised by the level above. From bottom to top, each group should let you say "therefore" before the sentence that sits over it.

Barbara Minto also offers a pattern for the opening sentences of a document, summed up in four words: situation, complication, question, answer. You recall a context your audience already knows, say what has changed or what is causing a problem, the question follows from it, and your answer is the top of the pyramid.

Out loud, at the end of a case, this fits in two sentences: "You asked whether the eight loss-making stores should be closed. My recommendation is to close five and renegotiate the other three." Recalling the question takes five seconds and guarantees that you are answering what was asked, which is not always the case after twenty-five minutes of analysis.

Here is the case. Ludibulle, a chain of 50 toy stores, is wondering whether to close its 8 loss-making stores. Over the course of the case, the candidate has established the following facts. The 8 stores lose €2M a year in total. Five are in city centres, lose €1.5M between them, and their leases end within the year: closing them would cost €1M. Three are in shopping centres, lose €0.5M between them, and their leases run for six more years: exiting would cost €2M in penalties.

Before. "So, I started by looking at the losses of each store, and we saw that the eight were losing €2M. Then I looked at the leases, there are five ending soon and three that run for another six years. After that we worked out the closing costs, €1M for some, €2M for the others. So I think we should maybe close some stores but not all of them."

Everything is accurate, and the interviewer waits forty seconds to learn that the answer is "some but not all", without knowing which.

After. "My recommendation is to close the five city-centre stores and keep the three shopping-centre stores while renegotiating their rents. Two reasons.

First, closing the five city-centre stores pays back in under a year: they lose €1.5M a year and closing them costs €1M, because their leases are expiring.

Second, closing the other three would destroy value: they only lose €0.5M a year, and exiting would cost €2M in penalties, four years of losses.

The main risk is losing the customers of the closed stores. The next step is therefore to check what share of their purchases would move to nearby stores and the website."

The facts are the same. The second version says what to do in the first sentence, and each number arrives to prove something.

The pyramid breaks when you do not have an answer yet. It shapes a conclusion; it does not replace the analysis. A candidate who announces "my recommendation is..." without having decided fills the top with an empty sentence such as "we need to optimise costs and grow revenue", and the rest of the pyramid supports nothing.

It also breaks when the top is contested in advance. Opening with an unwelcome conclusion to someone who is not prepared for it can put them on the defensive before they have heard the first argument. In that case, a construction that starts from shared observations and leads step by step to the conclusion lands better. In a case interview the question does not arise: the interviewer expects the answer first.

Third limit: the fake pyramid. Three headings under a title do not make three arguments. "Market, competition, company" is an outline, not a demonstration. Each box needs to contain a statement that someone could disagree with.

Finally, form lends confidence to fragile ideas. A well-structured recommendation looks solid even when one argument rests on an unverified assumption. A good synthesis says what it relies on and what remains to be confirmed.

Apply the pyramid to everything you say, not only the conclusion. When the interviewer asks what you read on a chart, start with what it shows ("the drop comes entirely from the professional segment"), then give the numbers. When they ask for ideas, announce the families before detailing them.

For the final synthesis, a four-part pattern works in almost every case.

1. The answer, in one sentence, using the words of the question.
2. Two or three arguments, each with the number that proves it.
3. The main risk, to show you can see the limits of your recommendation.
4. The next step, concrete.

Take ten seconds before speaking to write the top and the arguments on your sheet. It is the best-invested time of the interview.

Your opening structure prepares this work. If your issue tree split the question into three branches, your closing arguments are often the answers to those three branches. The FrameLab course stresses this link: the structure of the case and that of the conclusion are two sides of the same logic, and it devotes a whole module to synthesis. You can practise on the interview questions by forcing yourself to answer each one with the conclusion first.

A candidate ends a case about a long-distance coach company with these sentences, in this order.

• "We saw that the average load factor is 55%."
• "Routes to mid-sized towns are at 35% and lose money."
• "Routes between large cities are at 75% and are profitable."
• "A competitor has just stopped three routes to mid-sized towns."
• "So there is probably something to be done about the network."

Rewrite this synthesis as a pyramid. The client was asking how to return to profit.