BCG Casey is a chatbot-led online case that Boston Consulting Group uses to screen candidates, either before or alongside the interview rounds. It runs a compressed case conversation in which an automated interviewer presents a client situation and asks the candidate to answer multiple-choice questions, interpret charts, perform calculations and write short structured responses. There is no live interviewer, no opportunity to ask clarifying questions, and a running clock.
BCG’s own careers site is the authoritative source for what your particular application will involve, and the invitation you receive states the format for your sitting.
How it differs from a live case
The reasoning is the same. The conditions are not, and the differences change how you should prepare.
| Live case interview | BCG Casey | |
|---|---|---|
| Interviewer | A consultant who adapts to you | A scripted sequence |
| Clarifying questions | Expected and rewarded | Not available |
| Calculator | Not permitted | Generally permitted |
| Recovery from a wrong turn | The interviewer can redirect you | The case moves on |
| What is assessed | Reasoning, communication, presence | Reasoning and written communication |
| Pace | Conversational | Fixed and visible |
Two consequences follow. First, because you cannot ask what a term means, you have to read each prompt carefully and answer the question actually asked — candidates lose marks answering an adjacent question they found more interesting. Second, because the case does not adapt, an early misreading is not corrected for you. Reading precisely is worth more here than in a live case.
What the case asks for
The exercise follows the arc of an ordinary case: a client situation, a question about how to structure the problem, some quantitative work, one or more exhibits to interpret, and a closing recommendation.
Multiple-choice structuring questions. Typically a list of possible considerations, from which you select the ones relevant to the client’s question. The distractors are usually plausible-but-irrelevant rather than obviously wrong, so the discrimination is between “true about this industry” and “bears on this client’s decision”. Choose against the objective, not against general business sense.
Calculation questions. Standard case arithmetic — margins, break-even, growth, weighted averages — usually with the inputs supplied. A calculator is generally allowed, which removes the mental-arithmetic constraint but not the setup constraint. Choosing what to calculate remains the harder half, and it is where errors occur.
Chart interpretation. Exhibits with a question about what they show or what follows from them. This is the part most candidates find discriminating, because the charts are built so that a fast, careless read produces a specific wrong answer. Read the axis labels, the units and any footnote before extracting a value. Watch particularly for indexed values rather than absolutes, percentages of different bases, and stacked series where the segment of interest is not the one at the bottom.
Written responses. Short free-text answers, read and scored. These are the questions where structure is most visible and most rewarded.
Writing the free-text answers
Write them the way you would say them in a live case: the answer first, then the reasoning that supports it, then the caveat. Two to four sentences is usually right — the box is small and length is not the point.
A weak answer accumulates observations and leaves the reader to conclude:
“Revenue has grown but costs have grown faster. The premium segment is doing better than the standard segment. There are also some issues with the distribution channel that could be relevant.”
A strong answer commits, then supports:
“The margin decline is driven by cost growth rather than weak demand: revenue rose 12 per cent while total cost rose 19 per cent over the same period. Within cost, distribution accounts for most of the increase. I would want to see distribution cost per unit by channel before recommending a response, because the increase could reflect either rising rates or a shift towards a more expensive channel.”
That second answer states a conclusion, quantifies it, localises the driver, and names the evidence that would confirm it. It is also shorter than most candidates write.
Preparing for it
Practise chart reading against a clock. Take exhibits from case material, give yourself sixty seconds each, and write down what the chart shows and one thing it does not show. The second half of that drill is what catches over-reading, which is the main failure mode.
Drill answer-first writing. Take any case question you have worked through and write the answer in three sentences, conclusion first. Most candidates write conclusion-last by default, because that is the order they thought in.
Do full cases. The reasoning Casey tests is ordinary case reasoning, so the worked case studies remain the main preparation. The profitability walkthrough in particular covers the driver-isolation logic that these online cases lean on heavily.
Set the room up. Calculator, paper, pen, stable connection, no interruptions, and a browser you have not left twenty tabs open in. Candidates lose time to avoidable friction more often than they expect.
Do not rush the reading to save time for the answer. The clock invites this and it is a trap. A carefully read prompt answered in less detail beats a misread prompt answered thoroughly.
After the online case
Casey is a screen. Passing it leads to the live rounds, where BCG runs both case and fit interviews with consultants. The BCG firm profile covers that process and what the firm looks for, and the case interview hub sets out a full preparation sequence.
If you are also applying to McKinsey, note that its screen is a different design with different demands — the McKinsey Solve guide covers it. Bain and other firms use aptitude and behavioural assessments of another kind again, covered in the guide to consulting digital assessments.
