Case interview mistakes: find the cause and fix it.

A clinic wants to reduce the time patients wait for appointments. You build a structure about increasing clinic revenue. Higher revenue might eventually fund more staff, but you have changed the objective before establishing what is causing the delay.

Write the actual target at the top of your page. Confirm what success means and any constraint that changes the answer. In this example, that could be reducing the wait for a routine appointment within the existing staffing budget.

Then check your proposed branches against that target. Demand, available appointments and how appointments are allocated could each explain the wait. Advertising to attract more patients probably does not belong at the front of the investigation.

Practise this repair with short prompts: state the decision, the outcome and one question that would materially change your approach. Use clarifying questions when you need help deciding what to ask.

“Customers, competition, company and costs” gives you labels. It does not yet explain how you will decide whether a rural delivery service should add a new route.

Translate the labels into questions the client needs answered: How many deliveries will customers buy? What will they pay? How many stops can a driver complete? Will the route contribution cover its added costs? What happens to existing service?

You can still organise these into demand, economics and execution. The difference is that the questions now belong to the problem. See building a case framework for the full process.

A useful repair exercise is to take the same broad structure and adapt it to two very different businesses. If only the company name changes, keep working.

Suppose a subscription company loses money because customers leave before it recovers acquisition spending. You spend several minutes estimating the overall market, then request competitor advertising budgets. Neither analysis addresses the problem you have already found.

State what you are trying to learn before requesting data: “I want to see whether the early churn is concentrated in a particular acquisition channel. That would tell us whether to change targeting or investigate the experience after signup.”

This makes the request purposeful. It also helps you notice when the answer is sufficient to move on. The habit to practise is choosing and explaining the next analysis, not filling every box in your opening structure.

Revenue rises from $60 million to $75 million. The increase is $15 million. Dividing by $75 million gives 20%, but growth is measured against the original $60 million, so the correct increase is 25%.

More mental multiplication will not fix the denominator mistake. Write the formula in words first: change divided by starting value. Label the units and period, calculate, then check the scale.

A similar trap is comparing monthly revenue with annual costs. Another is treating a margin percentage as a profit amount. Before practising speed, identify whether the failure was setup, arithmetic or interpretation. Each needs a different exercise.

If you discover an error during the interview, correct it cleanly: “I used the ending revenue as the denominator. Growth is 25%. That makes the expansion larger than I first stated.” Then update any conclusion affected by the correction.

“The premium segment is growing and the basic segment is shrinking” describes a chart. It does not tell the client what to do.

Look for the implication. If premium orders contribute much more per order, the mix shift could explain stronger profit even while total order volume falls. Check the segment economics before claiming that it does.

Practise giving three things after an exhibit: the important finding, its business meaning and the next question. Keep the sequence short enough for the interviewer to follow. Our exhibit analysis guide develops this with worked examples.

You recommend automation because a $300,000 investment saves $120,000 a year, giving a 2.5-year payback. The client requires payback within three years.

The interviewer then explains that the saving is only $60,000 because the equipment will run for one shift. Payback becomes five years. Continuing to sell the original recommendation makes your confidence work against you.

A stronger response is direct: “At one shift, the investment misses the three-year requirement. I would not approve it on those economics. We would need either a lower investment or a credible way to increase the annual cash saving.”

Confidence includes revising your position. In practice, ask your partner to change one decisive assumption and observe whether you update the calculation, the recommendation and the next step.

“We looked at demand, costs and competition” tells the interviewer what you did. Finish by telling the client what you recommend and why.

For the automation example, the answer could be: “Do not invest under the current one-shift plan. Its five-year payback exceeds the three-year limit. A supplier offer at $180,000 or below would meet the limit at the stated $60,000 annual saving, so I would test that before reopening the decision.”

The condition is useful because it comes from the analysis. A generic “we should do more research” would leave the decision where it started.

Choose the earliest recurring breakdown. If you repeatedly miss the objective, start there. If the structure is sound but you lose direction after an exhibit, practise the transition from finding to next analysis.

Use a short repair loop:

1. Reconstruct the moment that went wrong, including what you said or wrote.
2. Make a second attempt at that moment with a clear correction.
3. Try the same skill on an unfamiliar prompt without help.
4. Bring it back into a full case and check whether it survives the added demands.

Keep the evidence in your practice log. “Needed a prompt to interpret the result” is more useful than a general score of seven out of ten.

For your next session, choose one problem from the case library and tell your partner the behaviour you want them to observe. Ask for the exact moment it worked or failed. That gives you a correction you can carry into tomorrow.