Strategic and case-style questions are not testing whether you know the answer. They test how you think when you are not sure.
Ten items. Work through the activities and say your answer out loud before you reveal it.
Three terms: first principles · lever · caveat. Say the right one, then reveal.
A composite scene, built for teaching: the people and figures are invented, but the pattern is one that interviewers describe again and again.
A senior advisory firm is hiring a partner. In the second round, each finalist is given the same open question: “A mid-sized healthcare group has had flat revenue for two years. How would you assess the problem and what would you recommend?” There is no data, no slide and no right answer.
The first finalist replies at once: “Marketing, almost certainly. They need to invest in digital channels, perhaps cut prices in the weaker regions, and probably restructure the sales team.” He speaks fluently for two minutes, with confidence, using three recommendations that sound reasonable.
The panel notices what is missing: no question about what “flat” means, no idea of how he reached marketing, and no indication of what he would do if he was wrong.
1. What was the question, and what information did the finalists have?
2. What three things did the panel notice were missing from the first answer?
1. A mid-sized healthcare group with flat revenue for two years: how to assess it and what to recommend; there was no data or slide.
2. No clarifying question about what “flat” meant, no reasoning behind marketing as the cause, and no plan for being wrong.
The second finalist pauses. She says, “That’s a good question. Let me take a moment to structure it.” She is silent for about ten seconds, which feels long to everyone, including the panel. Then she asks three things: is revenue flat in every region and service, or only in some? is the goal growth or profit? and what has changed in the market in the last two years?
Having received short answers, she says: “I’d look at it in three parts: the market, the group’s offer, and its ability to execute. My working hypothesis is that one segment has matured while the group has not invested in new ones. I’d test that first, because it is the quickest to confirm or reject.”
She recommends two small pilots in adjacent segments, names the main risk — distraction from the core business — and says what would make her change course: less than ten per cent uptake after six months. Her answer is not obviously better in content, and the panel does not know whether the hypothesis is right. But they know exactly how she thinks.
1. What did the second finalist do before she answered, and how long did she take?
2. What did her recommendation include that the first answer did not?
1. She paused for about ten seconds to structure her thinking, then asked three clarifying questions about scope, goal and market change.
2. A stated hypothesis, a way to test it, a named risk and a clear signal for changing course.
A panel member now says to the second finalist, flatly: “I don’t agree with your hypothesis.” It is not clear if he means it or is testing her. She does not defend and she does not retreat: “That’s helpful. Which part do you disagree with? If it’s the assumption about demand, I’d be happy to revisit it. If new data showed the segment was larger than I assumed, my recommendation would change.”
He then gives a fact that undermines her assumption: the group’s newest segment has been shrinking. She takes a breath, rebuilds her structure in front of them, and revises her hypothesis in three sentences.
The panel offers her the role. A year later, one member admits that the first finalist may well have been right about marketing. Nobody can say, and the panel is honest that it chose the person whose thinking it trusted, not the answer it agreed with.
1. How did she respond to “I don’t agree”?
2. What did the panel say about its own decision a year later?
1. She asked which part he disagreed with, offered to revisit the assumption, and said what evidence would change her recommendation; when he gave new facts, she rebuilt her answer.
2. That the first finalist may have been right about marketing; the panel chose the thinking it trusted, not the answer it agreed with.
Case answers are made of uncertain statements. The grammar of uncertainty — modals, conditionals and hedges — is what lets you sound decisive and honest at the same time.
Complete each sentence with one word.
“Quality scores are stable, so it be a product problem.”
“Revenue is flat in only one region, so it well be a regional issue.”
“Both competitors cut prices in the same month, so that be the main cause.”
“A price rise is to reduce volume, but it is not certain.”
can’t · may · must · likely
Can’t / must are logical deductions (not obligations). May well / likely to / could rank from probable to possible.
Put the verb in brackets into the correct form, or add the missing word.
“If demand to fall further, we would need to cut costs.”
“If the two pilots , we will extend them to a third segment.”
“ adoption exceeds ten per cent after six months, I would stop the pilots.”
“If I were advising the board, I recommend a smaller first step.”
were · succeed · Unless · would
“Were to” is used for a hypothetical, formal condition. After “if” in a real future condition, use the present simple, not “will”.
An interviewer asks how you would assess a stalled project. Complete the answer.
“Before I answer, may I two things? Is the problem across the whole portfolio or only some areas? I’d look at it in parts: the market, the offer, and execution. My working is that one segment has matured. On , I’d recommend a small pilot, with one : it assumes demand elsewhere is stable.”
clarify · three · hypothesis · balance · caveat
Notice the verbal map: clarify → three parts → hypothesis → recommendation → caveat. Each step makes the answer easier to follow.
Take these into the Speak tab: you will use exactly this language under time pressure.
This is a composite scene, so there is no real data. What follows is the argument about what a case-style question tests.
Most senior panels score on the same five things, whatever the question. You will use this scorecard to mark your own mock interview in the Speak tab.
| Criterion | What the panel is listening for |
|---|---|
| Structure | The listener can follow the shape of the answer (headline, steps, result). |
| Evidence | A specific example, with a number, instead of a general claim. |
| Ownership | It is clear what I personally decided or did, and what the team did. |
| Relevance | It connects to the role and the organisation I am talking to. |
| Delivery | Calm pace, no long fillers, finished within the time. |
A case question tests four behaviours at once: whether you clarify before solving, whether you structure so that the listener can follow, whether you can commit to a hypothesis, and whether you update when you are wrong. Content knowledge matters far less than panels admit.
A ten-second pause feels like an hour to the person pausing, and like a deliberate act to the listener. Say the first sentence that tells them you are thinking (“Let me structure that”), and then actually do it. Candidates who fill silence with the first idea usually answer a different question from the one asked.
The principled reading: the panel rewarded transparent reasoning, which is what a partner needs in front of a client. The sceptical reading: the panel rewarded a recognisable ritual — clarify, structure, hypothesise — which is a performance anyone can learn. A senior candidate uses the ritual and also has something real to say.
All of it out loud, against the clock. These are the pressure situations in which a senior person has to think aloud: the open case, the curveball, the challenge and the “I don’t know”.
Decide before you click, and say why.
A case answer has three steps. Write your own words for each, then compare with a model.
“A mid-sized organisation has had flat growth for two years. How would you assess the problem and what would you recommend?”
Clarify: “Before I start, may I ask three things? Is growth flat across all products or just some? Is the goal revenue or profit? And what has changed in the market in the last two years?”
Structure: “I’d look at this in three parts: the market, the organisation’s offer, and its ability to execute. My hypothesis is that one segment has matured while the organisation has not invested in new ones.”
Recommend: “On balance, I’d recommend testing two adjacent segments with small pilots over six months. The main risk is distraction from the core business, which I’d mitigate with a dedicated team and clear stop criteria. If the pilots show less than a 10% uptake, I’d change course.”
Notice: questions first, a visible structure, one hypothesis, a decision and a stated risk.
Close your notes and give the answer aloud in two to three minutes. Then give it again with one improvement.
Fill in your own details, then answer whatever the cards give you. Draw a question, answer it, then draw a follow-up.
Choose one question and estimate the answer aloud, with the logic visible. Do not use a calculator; round the numbers.
Structure: I’d estimate it in three steps. First… Second… That gives roughly… which I’d sense-check against…
A panel member says: “I’m not convinced. Why shouldn’t we simply do nothing?”
Answer properly: take the question seriously, state the cost of doing nothing, name what would make you agree with the panel, and close with a position. Then do it again in ninety seconds.
Six questions in a random order, as in a real interview. Use the timer for each answer, give each one two minutes at most, and do not stop to correct yourself. Score yourself at the end.
Choose the three you find hardest. Prepare a structure for each, then answer them aloud.