MECE: definition, examples and exercises with answers

MECE is the acronym of "mutually exclusive, collectively exhaustive", pronounced "me-see". The term is credited to Barbara Minto, who formulated it at McKinsey in the 1960s and made it one of the foundations of her pyramid principle.

The two halves are checked separately.

• Mutually exclusive: no item can be filed under two categories at once. If you split a shop's customers into "young people", "families" and "loyal customers", a loyal young parent falls into all three boxes. You will count the same effect three times.
• Collectively exhaustive: no part of the subject is left out. If you explain a drop in profit by "prices" and "costs", you have forgotten volume sold, and that may be where everything is decided.

The simplest example is age: "under 30, 30 to 60, over 60" is MECE. "Students, working people, over 60s" is not, because a 62-year-old worker is in two boxes and a 40-year-old job seeker is in none.

Some breakdowns are MECE without effort, because their logic rules out overlaps and gaps. Knowing five is enough for most situations.

• A formula. Profit = revenue - costs; revenue = price x volume. Any arithmetic breakdown is MECE, which is why the profit tree is so robust.
• A process. The steps of a flow, in order: order, preparation, shipping, delivery. Every minute of delay belongs to one step and only one.
• A segmentation on a single criterion. By age band, by region, by sales channel. The trap is mixing two criteria in the same list.
• An opposition. Internal and external, supply and demand, existing and new customers, short term and long term. A thing and its opposite cover everything.
• A timeline. Before, during, after; or year by year.

When your list fits none of these moulds, adding an "other" category makes it exhaustive. That is acceptable as long as it stays small. If "other" weighs a third of the total, your breakdown is not the right one.

The principle applies level by level. In the tree below, which breaks down a bookshop's sales, each node is split into parts that do not overlap and whose sum or product gives back the node. You can check by picking a transaction at random: it comes either from a card holder or from an occasional customer, never from both.

Each level can use a different way of splitting, a formula at the top, a segmentation below. What is not allowed is mixing two logics on the same level.

Here is the prompt. An independent bookshop saw its sales go from €500,000 to €440,000 in a year. Why?

Before. A candidate suggests looking at five leads: competition from online sales, fewer customers, prices, the children's section and the weather. The list is not absurd, but it is not MECE. Online sales and the weather are possible causes of having fewer customers, so they overlap with it. The children's section is a segment, not a cause. And nothing covers the case where customers come just as often but buy fewer books.

After. You start again from a formula. Sales = number of transactions x average basket. The interviewer gives the numbers: 25,000 transactions at €20 last year, 22,000 transactions at €20 this year. The basket has not moved. The whole drop comes from the number of transactions, which is down 12%.

You then split transactions on a single criterion, the type of customer. Loyalty card holders stayed at 15,000 transactions. Occasional customers fell from 10,000 to 7,000.

Two MECE levels were enough to rule out price, basket and loyalty, and to ask a precise question: why are passers-by coming in less? The interviewer's answer, roadworks blocking the street for eight months, was in none of the five initial leads. That is the point of exhaustiveness: it leaves room for a cause you had not imagined.

MECE breaks first because it says nothing about usefulness. Sorting a bank's customers alphabetically is beyond reproach and moves no problem forward. The right breakdown is the one that isolates where the problem concentrates, and that requires understanding the business, not applying a rule of logic.

It also breaks when causes interact. In reality, a price increase lowers volume, and a poor-quality product weighs on both sales and after-sales costs. Filing these effects in sealed boxes is a useful simplification to make progress, provided you put the pieces back together when you conclude.

Third limit: the obsession with perfection. In an interview, spending a minute working out whether "marketing" and "sales" overlap slightly costs you more than it earns. Rigour needs to be strong on the first level of your structure, where a gap would make you miss the subject, and can relax below.

Finally, MECE does not generate ideas. For a creative question, such as ways to cut queues at an airport, start by thinking freely, then sort your ideas into a MECE structure. Doing it the other way round stifles imagination.

Use MECE as a quality check that you run in ten seconds after laying out your structure. Two questions are enough.

1. Could the same item fall into two of my branches? If so, redefine the branches or pick a single splitting criterion.
2. What could explain the problem and is in none of the branches? If you find something, add it or widen a branch.

Then ask the question MECE does not ask: would this structure be different for another case? If you could reuse it as is for any company, it is too generic, even if it is logically perfect.

Say out loud what makes your breakdown complete. "Sales are the number of transactions times the basket, so by looking at both I cover everything." The interviewer does not have to check it themselves, and they hear that you are structuring deliberately.

The principle does not stop at the opening structure. It applies to a list of ideas in a brainstorm, to the arguments of a recommendation and to the segments of a market sizing. In the FrameLab course, MECE is presented as the test that validates a structure built with the five reasoning building blocks, never as the goal. The interview questions give you prompts to practise out loud.

Answer each question before opening the answers.

1. A bank splits its revenue for the year into three parts: new customers, existing customers, customers who left during the year. Is this breakdown MECE?
2. Suggest a MECE breakdown of a restaurant's costs.
3. Subscribers to a video streaming service are cancelling more and more. Give a first MECE structure of the possible reasons.