What Case Interviewers Score: Four Published Lists

The four firms that publish a case interview criteria list

Start with the fact that almost no guide mentions. These lists exist, in public, on the firms' own career pages, and you can read them before your interview.

Oliver Wyman publishes six criteria under its case interview section at oliverwyman.com. In our summary of the page, they cover how you approach an unstructured problem, how you evaluate and use data, whether you can think outside a given frame, the common sense you apply, the questions you ask, and how you communicate ideas. The page also tells you the stance it wants: "Think of your case interviewer as your client. They want you to solve the problem, and can help you to do so. Work together."

Accenture packs four criteria into one published sentence in its case workbook. Success "will not depend on identifying the "correct" answer, but rather on how clearly you define the problem, how logically you structure your analysis, your quantitative analysis skills and how well you communicate your thoughts to the interviewer." Problem definition, structural logic, quantitative skill, communication. The same document adds a set it labels soft evaluative: Poise, Communication, Flexibility, plus energy, initiative, time management and decisiveness. It is published at accenture.com, carries a 2018 copyright, and that date matters when you read it.

Deloitte publishes five dimensions inside its own practice tool, each with an instruction attached: "Clear thinking", "Strategic framing", "Practical judgement", "Summary analysis", "Final recommendations". The spelling of the third is the firm's own, and the list sits inside the practice tool at usrecruiting.deloitte.com, which the firm states is not an assessment and does not record answers.

Simon-Kucher names five competencies it assesses consistently, and separately five performance factors on the case itself. One of its competencies has no counterpart anywhere else: the ability to work with AI-supported tools and to critically evaluate what they produce, published at simon-kucher.com.

Where the four lists agree, and the one item only Accenture names

Set the four side by side and three things appear in all of them.

Three columns of that table are the reason a case interview feels the way it does. Structure, arithmetic and speech get assessed at once, continuously, from the first minute.

Two rows deserve attention because they are thin. Only two of the four firms credit the questions you ask, and the two disagree about how far to push. Deloitte writes: "Ask questions, but don't use them as a crutch. While it's important to ask clarifying questions, avoid relying on them too heavily. We're interested in how you connect the dots when solving the problem." L.E.K., on a separate page, makes asking questions an entire numbered step of its own method. Both are published. The equilibrium point differs by firm, and that is worth knowing before you open with five clarifying questions.

The composure row has exactly one entry. Accenture names Poise and Flexibility, and no other firm in our corpus names either.

The two-tier case interview ladder Bain and Deloitte publish

Two firms publish something more useful than a criteria list. They publish worked cases where each question carries two graded answers.

Bain structures both of its home-grown cases as "A good answer" then "A better answer". On the coffee shop breakeven question, the good answer produces the number: 409,350 pounds of cost over a 2 pound unit margin gives 204,675 cups, which the page compares to a 7,000,000 cup market and calls roughly 3 percent of share. We recomputed that division by hand on 30 September 2026 and it holds: 204,675 divided by 7,000,000 is 2.92 percent. The better answer adds something the number does not contain. Bain writes that it "would take into account how long it would take for her to recoup the cost to open the store and other considerations before proceeding", quoted from bain.com.

Deloitte's LeadAuto case uses the same device with different wording. It separates "A good answer may include" from "In addition to the above, a great answer may also include", and the upper tier adds an internal and external split, a multi-year contract commitment, a joint venture, expansion beyond Asia and higher-margin services.

Two firms, no connection between them, the same mechanic. And the content of the upper tier is the same kind of material in both: risk, alternatives, time horizon, next step. Two instances do not make a market pattern, so we report it as what it is. Two firms describe the same gap, in their own published answers, and that gap is where the second half of the credit sits.

Our weights, and the thing they cannot see

No firm publishes a weight. That holds across all forty-four research files behind this site, and it is the most stable fact in the corpus. The numbers below are ours, read from the grading code in src/lib/grading, and no consulting firm states anything like them.

Reading a published criterion against one of our dimensions

When we line up a firm's published words with one of our weights, that is our reading. Neither firm states an equivalence, and the boundaries differ.

1. Oliver Wyman on unstructured problems reads against Coverage and Separation, 70 of 100 together. Its criterion asks how you approach a problem with no given frame, which is what those two lines measure between them.
2. Deloitte's "Strategic framing" spans two of our dimensions at once. Its instruction is to "Develop a logical structure and identify assumptions; take notes while the scenario is being read". Structure and assumptions sit in separate rubrics for us. The correspondence is partial, and calling it exact would be wrong.
3. The Bain and Deloitte upper tier maps onto Risk 20 and Next step 20 in our synthesis rubric. That is 40 points of 100, and both firms put exactly that material in the higher tier of their own answers. It is the closest thing to independent corroboration we have for a weight.
4. Accenture's Poise maps onto nothing of ours. Stated plainly, because a page that claims full coverage of a published list it cannot score would be selling something.

The practical version of all this: the four lists tell you what to do, and our weights tell you what it costs when you skip one. Neither replaces the other.

What none of them publishes about a case interview, and what to do

Six absences, and they are consistent across every firm we have read.

• No weights. Zero, across forty-four firm files.
• No pass threshold. No minimum score, no number of criteria you must satisfy, no count of cases you must clear in a round.
• No aggregation rule. Bain writes that feedback rests on "an objective scoring system" and describes not a single element of it.
• No published duration for a case interview at McKinsey, Bain, Kearney, L.E.K. or Simon-Kucher. Accenture's workbook gives 45 to 60 minutes total, with 30 to 45 on the case, and that document carries a 2018 copyright.
• No definition of Poise. Accenture names the criterion and leaves it undefined.
• No confirmation that the six Oliver Wyman criteria apply outside the case. Its page places them under the case interview. Its second interview type has no published criteria at all.

One published detail cuts against the whole idea of a fixed list, and it deserves the last word. Oliver Wyman's Dairy Farm case prints evaluation criteria under each individual question, and they change from question to question. At the opening it wants the key question identified, a clear structure, and that structure communicated succinctly. On the following question it wants several factors produced, adaptation as data arrives, and the data laid out for fast computation. Same case, same candidate, different standard two minutes later.

Turning the four lists into one case interview practice session

Here is how to use the material above without a subscription to anything.

1. Read the four lists on the firms' own pages. They take about fifteen minutes in total. Two of the four sit behind JavaScript, so read Deloitte's dimensions inside its practice tool rather than on its tips page.
2. Run one case, then grade it against six items. Problem definition, structure, arithmetic, questions asked, communication, composure. That covers every distinct theme the four lists name.
3. Apply the two-tier test to your own recommendation. Write your answer down. Then ask whether it contains a risk, an alternative, a time horizon and a next step. If not, you produced the lower tier of Bain's published ladder.
4. Rehearse the opening twice. Coverage is 45 points in our rubric and the opening is where it is won or lost.
5. Time yourself against 30 minutes. That is the lower bound Accenture publishes for the case portion, and it is the only published number of its kind.

Our structure drill reports Coverage, Separation, Depth and Concision as separate lines with the first correction named, and our synthesis drill reports Answer first, Evidence, Risk and Next step. Both are free and need no account. Neither scores composure, because a text grader cannot hear it, so that one stays on your list of things to practice with a person. For the firm-by-firm detail behind the lists above, see the Oliver Wyman case interview guide and the Deloitte case interview guide. For the mechanics of how a score gets built during the conversation, see how cases are scored.