An interviewer asking “Are you sure?” has not told you that you are wrong. They have asked you to examine your reasoning.
Listen for what is being challenged. Is it a calculation, an assumption, the evidence, the choice of next step or the feasibility of your recommendation? Respond to that issue directly. You will sound more composed when your attention is on the problem rather than on whether the interviewer approves of you.
My default would be a brief pause, a clear response and a next move. Long apologies and defensive explanations rarely improve the analysis.
Find the question inside the challenge
Different challenges need different answers:
| What the interviewer says | What to do |
|---|---|
| “Why start there?” | Explain what that analysis could change |
| “What supports that conclusion?” | Point to the evidence and connect it to the claim |
| “Could something else explain it?” | Compare your explanation with the alternative |
| “The client cannot do that.” | Apply the new constraint and redesign the option |
| “Check your calculation.” | Rebuild the relevant step and update its consequence |
Treat the wording as information. “Why?” is an invitation to explain; “the site cannot open before December” is a new fact. You should not respond to both by instantly reversing your view.
Defend the reasoning when it still holds
Imagine a fictional service that has passed its acceptance standard on routine work. Complex work has not passed. You recommend a routine-only rollout.
The interviewer asks: “Doesn’t the poor complex-task result mean the whole trial failed?”
A useful response is:
“I would keep the routine-task recommendation. That segment meets the required standard; the new concern is specific to complex work. I would check whether the failure mechanism also appears in routine tasks before expanding the restriction.”
You have answered the challenge without dismissing it. The reasoning gives a condition under which the broader concern would matter.
Avoid relying on confidence alone: “I still believe it is a good idea” does not explain why. Keep the view because the supporting evidence survives the challenge.
Revise the part that new evidence overturns
Now change the situation. You originally recommended deploying across both task groups because an overall acceptance score cleared the threshold. The interviewer reveals that routine work dominates the average while complex work performs poorly.
The correct response is to narrow the deployment:
“That split changes the scope of my recommendation. I would deploy only for routine tasks and keep complex tasks on the existing process until they meet the required standard.”
There is no need to rescue the original all-workload recommendation. The new evidence is more informative than the aggregate score, so use it.
Recalculate the total from each group’s volume and rate. A shift towards harder work can worsen the overall result even when each group improves.
Quick Business Judgment Drill
You recommended that an energy supplier's new chatbot handle all customer queries. The interviewer then shows its results split by query type.
- Required resolution rate
- 80%
- Overall resolution rate
- 82%
| Query type | Monthly queries | Resolution rate |
|---|---|---|
| Billing questions | 9,000 | 88% |
| Fault reports | 3,000 | 64% |
Given this split, how should you revise your chatbot recommendation?
Correct an error all the way through
Suppose you say an investment of £450,000 with annual net savings of £90,000 has a four-year payback. The interviewer asks you to check.
The division is five years. If the client requires four years, the proposal misses the hurdle. Say both:
“I made an error in the division. Payback is five years, so the current proposal misses the four-year requirement. To meet it, annual net savings need to rise to £112,500, or the investment needs to fall to £360,000.”
The correction is complete because it changes the recommendation and identifies the gap. Keeping the old recommendation after correcting the number would leave the substantive mistake in place.
You need not apologise repeatedly. Acknowledge the error once, show the repaired reasoning and continue.
Use a rival explanation to improve the next test
Suppose insurance claim costs have risen and you suggest repair inflation. The interviewer asks whether the insurer’s claims have shifted towards more expensive vehicle types.
A useful response is:
“That could also raise the average. I would compare repair cost within vehicle categories and then examine the change in claim mix. If within-category costs are stable, I would shift the explanation towards mix.”
This response turns the challenge into a test. It also avoids pretending that your first explanation was established fact.
Use a good alternative to sharpen your analysis. Building on another person's idea is part of solving the case together.
Quick Answer Repair Drill
A parcel firm's cost per delivery has risen, and you blamed higher fuel prices. The interviewer asks whether more deliveries now go to rural addresses.
- Your reply
- "Possibly, but fuel is clearly the main driver."
What is the strongest repair to this reply?
Handle a hard constraint without trying to negotiate it away
If the interviewer says the fleet cannot grow this year, proposing more vehicles again does not show persistence. Work inside the constraint: schedules, turnaround time, route allocation, maintenance availability or demand timing may still be adjustable.
State the consequence before generating alternatives:
“With fleet size fixed, I would focus on usable vehicle hours and how they are allocated across routes.”
A constraint can make brainstorming more productive by identifying the levers that remain available. For a transit operator with no extra vehicles, for example, test dispatch timing, turnaround delays and route allocation before proposing fleet expansion.
Avoid reading every cue as a verdict
A repeated question may mean your explanation was unclear. A new exhibit may be the planned next stage. A challenge may be testing whether your view has support. You cannot reliably infer the final interview outcome from one facial expression or follow-up.
Focus on the actionable content: what needs explaining, checking or changing? If you did not understand the question, ask for it to be repeated.
Practise with three different kinds of challenge
Ask a partner to give you one request for justification, one new constraint and one correction. Your response should differ in each case. Explain sound reasoning, revise an affected decision and repair an actual error.
Try Auto Insurance Claim Severity. Give a working explanation, then consider the strongest alternative. If the challenge leaves you unsure where to go next, restate what is known and identify one comparison that would distinguish the remaining explanations.
Point of view drill
Would your argument persuade an MBB interviewer?
Auto Insurance Claim Severity
Use an insurer’s frequency and severity evidence with Casey, Case Room’s AI interviewer. Distinguish a supported explanation from a cause the data has not established.
Frequently asked questions
Working out what is being challenged
Does ‘Are you sure?’ mean my answer is wrong?
No. It asks you to examine your reasoning, not to abandon it. Work out what is being questioned: a calculation, an assumption, the evidence, your chosen next step or the feasibility of the recommendation. Then respond to that specific issue. This keeps your attention on the problem rather than on whether the interviewer approves of you.
Should I answer ‘Why start there?’ the same way as ‘What supports that conclusion?’
No, because they ask for different things. ‘Why start there?’ asks what that analysis could change about the decision. ‘What supports that conclusion?’ asks you to point to specific evidence and show how it connects to your claim. If you answer a sequencing question with evidence, or the reverse, the real challenge goes unaddressed.
If the interviewer repeats a question, does that mean I am failing?
Not necessarily. A repeated question may simply mean your explanation was unclear, so restate it more directly. A new exhibit may be the planned next stage, and a challenge may test whether your view has support. You cannot reliably infer the outcome from one follow-up or facial expression. Focus on what needs explaining, checking or changing.
Can I ask the interviewer to clarify a challenge I do not understand?
Yes. If you are unsure what is being questioned, ask for clarification rather than guessing and defending the wrong point. A short request is more useful than a long answer to a question nobody asked. It also lets you respond to the actual issue, whether that is a number, an assumption, the evidence or the next step.
Deciding whether to hold or change your view
If my point still holds, should I change my answer anyway to seem open-minded?
No. Keep the recommendation and explain why it still holds. In the guide's example, a service meets the standard on routine work but not on complex work. A routine-only rollout therefore survives the question of whether the poor complex-task result means the whole trial failed. The concern is specific to complex work, which your recommendation already excludes.
How do I know when new evidence should change my recommendation?
Revise when the evidence removes the basis for part of your conclusion. Suppose you recommended deployment across both task groups because an overall score cleared the threshold. If you then learn that routine work dominates the average and complex work performs poorly, that basis no longer holds. Narrow the scope to routine tasks and leave complex tasks unchanged, rather than discarding everything.
What should I do when the interviewer says a constraint cannot change?
Accept the constraint and work inside it. If the fleet cannot grow this year, proposing more vehicles again does not show persistence. First state the consequence, for example that you will focus on usable vehicle hours and how they are allocated. Then look at what can still be adjusted: schedules, turnaround time, route allocation, maintenance availability or demand timing.
Fixing errors and testing alternatives
What if I am challenged on a calculation I am confident is correct?
Check it visibly, confirm that it holds and continue. ‘Are you sure?’ asks you to examine your reasoning. It is not proof of an error. Briefly restate the inputs and the operation, confirm the result and connect it back to the decision. The conversation then moves forward rather than stalling while you seek reassurance.
I made a calculation error. Is giving the corrected number enough?
No. Carry the correction through to the decision. Suppose an investment of £450,000 with £90,000 annual net savings pays back in five years, not four, and the client requires four. Say that the proposal now misses the hurdle. Then state what would need to change, such as annual savings of £112,500 to recover the investment within four years.
What if the interviewer suggests a different explanation from mine?
Treat it as a rival hypothesis and design a test that separates the two explanations. Suppose you blamed rising claim costs on repair inflation and the interviewer suggests a shift towards expensive vehicles. Compare repair cost within each vehicle category, then examine the change in claim mix. If costs within categories are stable, the explanation moves towards mix.
Practising under pressure
How can I practise handling pushback with a partner?
Ask your partner for three different challenges: a request for justification, a new constraint and a correction. Each response should be different. Explain reasoning that is sound, revise the decision the constraint affects and repair the genuine error. The guide also suggests the Auto Insurance Claim Severity drill. In it, you give a working explanation and then consider the strongest alternative.
What should I do if a challenge leaves me unsure where to go next?
Restate what is already known, then identify what remains open. This gives you and the interviewer a shared starting point. It also turns uncertainty into a specific next step, such as a figure to check or an alternative to test. That is more useful than guessing or repeating your previous answer more confidently.