Structured Thinking 2026: Three Firms, Three Rules

Three published instructions on structured thinking that disagree

Start with the words, all three checked 30 September 2026 on the firm's own page.

Kearney goes further in the same paragraph. It names what a structure does have to fit: "The only things it needs to fit are the problem, the business, and the objectives." That is at kearney.com, read through a text proxy because the site refuses our direct requests.

Simon-Kucher's line is signed. Michaela Scherhag, Team Lead Junior Recruiting in Frankfurt, wrote it on the firm's own careers page at simon-kucher.com. She names pricing first among the areas where knowing the frameworks pays.

So a candidate who reads two firms gets two orders. One says the standard shape is a risk. The other says the standard shape is an asset, at least in pricing. Both are the firm speaking about its own interviews. Neither mentions the other.

What all three firms actually ask of a case interview structure

Read the three lines again and the disagreement narrows. None of them says whether to have branches. All three assume you will. What they disagree about is where the branches come from.

Kearney wants them to come from the problem in front of you. Simon-Kucher wants you to know the standard shapes well enough to reach for the right one. Bain wants them to move as the conversation gives you new facts. Those are three answers to a single question: what generates the split.

There is one answer that satisfies all three at once. Write the equation of the thing you were asked about, then read the branches off the equation. A structure built that way fits the problem by construction, which is Kearney's test. It will often land on a shape with a name, which satisfies Simon-Kucher. And it updates cleanly, because a new fact changes a term rather than the whole drawing, which is Bain's point about building on ideas.

This also fixes the failure mode that the firms describe without naming. A recited shape has branches that were chosen before the problem was read. They may still be the right branches. You cannot tell, and neither can the interviewer, and that is what loses the points. Our page on case interview frameworks works through the named shapes one by one; this one is about where any shape has to come from.

Branches read off an equation, with numbers

Take a subscription business whose revenue fell. Revenue equals subscribers times price. Subscribers at the end of a year equal subscribers at the start, plus new, minus churned. That is the whole equation, and it hands you the branches.

Here are the figures for the year, which you would ask for rather than assume.

1. Start of year: 480,000 subscribers at 25 dollars a month.
2. Added during the year: 96,000 subscribers, which is 20 percent of the opening base.
3. Churned during the year: 144,000 subscribers, which is 30 percent of the opening base.
4. End of year: 480,000 plus 96,000 minus 144,000, so 432,000 subscribers at 24 dollars a month.

Now the arithmetic, done in the order the equation gives.

Opening run rate: 480,000 times 25 dollars is 12,000,000 dollars a month, and times 12 months that is 144,000,000 dollars a year.

Closing run rate: 432,000 times 24 dollars is 10,368,000 dollars a month, and times 12 that is 124,416,000 dollars a year.

The decline is 19,584,000 dollars, which is 13.6 percent of the opening run rate.

Two branches, and here is where the case gets interesting. Attribute volume first, at the old price: the loss of 48,000 subscribers at 25 dollars over 12 months is 14,400,000 dollars. Then price, at the new volume: 432,000 subscribers losing 1 dollar a month for 12 months is 5,184,000 dollars. The two add to 19,584,000, which is the whole decline.

Reverse the order. Price first, at the old volume: 480,000 times 1 dollar times 12 is 5,760,000 dollars. Then volume, at the new price: 48,000 times 24 dollars times 12 is 13,824,000 dollars. These also add to 19,584,000.

Both decompositions are arithmetically sound and they disagree about how big the price problem is. Price carries 5,184,000 on one ordering and 5,760,000 on the other, so between 26.5 and 29.4 percent of the decline. The number you report depends on a convention, and saying which convention you used is the whole difference between a structure and a guess.

The branch that double counts, and how to spot it

Candidates who know this equation still lose points by adding a third branch that looks new. Plan downgrades is the usual one. It feels like a distinct driver of the decline, and it is already inside the price branch, because average price is what a mix shift moves.

The test takes five seconds. Add the branches up and see whether the total exceeds the thing you were asked to explain. Price plus volume already accounts for 19,584,000 dollars of decline. Any third branch at the same level has to be worth zero, or two of your branches are describing the same dollars.

That test is cheaper than any mnemonic about mutual exclusivity, and it generalizes. A branch belongs at a level when it is a term in the equation at that level. A plan downgrade is not a term; it is a cause of a term, so it sits one level down, under price, with mix and discounting beside it.

Oliver Wyman publishes a related instruction under its own structuring step, checked 30 September 2026: "Show your client why it's better than the alternatives. State your assumptions." An equation makes both cheap to do. The alternative structures are the other ways to split the same total, and the assumptions are the attribution order and the base.

Holding a line and changing it in the same case interview

L.E.K. publishes seven numbered steps for a case at lek.com, checked 30 September 2026. Step 3 is titled "Stick to your approach", and it says to "Outline how you are thinking and move methodically to address each issue, point-by-point". Step 6 says to "Commit yourself to a course of action, but show flexibility and a willingness to rethink in the face of new data".

Three steps apart, one firm, two instructions that pull against each other. Kearney does the same thing inside one sentence: "Be coachable, soliciting feedback and integrating it quickly. At the same time, be firm about the things you believe in."

Neither firm resolves it, and the conflict comes from the exercise itself. An equation resolves it in practice, because it separates the two things a candidate is asked to hold. The structure is the equation. The estimates in it are the numbers. New data revises a number, and the equation stands. New data that adds a term revises the equation, and you say so out loud when it happens.

That gives you a sentence for each side. Holding the line sounds like this: the split is still volume and price, because revenue is still the product of the two. Changing sounds like this: churn turns out to be concentrated in one cohort, so the volume branch needs a cohort term, and I am adding it now. Both are the same posture, applied to different objects.

How we score a structure, and who wrote the scale

The weights below are ours, read from the grading code in src/lib/grading. No consulting firm publishes a marking scheme, on structure or anything else, and that absence is one of the steadiest findings in our research files.

Drilling the derivation of structured thinking rather than the shape

Practice the derivation, because that is the part that transfers. Take any prompt and write the equation before the branches, then check that your branches are its terms.

Three drills that work without a partner, using the numbers above.

1. Change the base and redo the split. Set opening price to 25 dollars and closing price to 22.50. Closing run rate becomes 432,000 times 22.50 times 12, which is 116,640,000 dollars. The decline is 27,360,000 dollars, and price now carries 480,000 times 2.50 times 12, or 14,400,000 dollars, on the price-first ordering. Price has gone from about a quarter of the problem to more than half.
2. Solve backwards. Ask what churn would have held revenue flat at 144,000,000 dollars with price at 24 dollars. You need 144,000,000 divided by 288 dollars a year, so 500,000 subscribers, which is 20,000 above the opening base. At 96,000 additions, churn would have to fall to 76,000, or 15.8 percent of the opening base, rounded.
3. Break your own structure. Name a fact that would force a new term, then write the revised equation in one line.

Run the same habit against a timer in our structuring drill, which returns Coverage, Separation, Depth and Concision separately with the first correction named. The math drill times the arithmetic this page leans on. Both are free and need no account. Neither will tell you which equation the prompt deserves, and that judgment is the thing being interviewed.

When the prompt gives you nothing to write an equation about, the move is different, and unstructured cases covers it. When the question you need answered before structuring is missing, see clarifying questions.