Michigan Ross Case Book: Its Board Signed a Warning

What the Ross board wrote about its own case book

We read the letter, the acknowledgements and the format note, checked 30 September 2026, from a copy hosted by the career service of an unrelated university. The signature reads "2019-2020 Board, Consulting Club @ Ross". Seventeen board members are named, one of them holding a post called VP Casebook, and thirteen case writers are named separately, each credited with a specific case.

The provenance sentence is first: "In order to facilitate your preparation, your fellow club members have recorded their real-life case interview experiences and their customized frameworks and solution elements."

The warning follows immediately, in the same paragraph: "These cases act as a strong reference point for what to expect during a consulting interview but are in no way all-encompassing. Each case comes down to a conversation between the interviewer and the candidate, so it is very possible that two candidates could have two very different conversations about the same business problem. In fact, we encourage this."

On where the problems come from: "These cases, or mini-business problems, are a glimpse into a consultant's (and often the interviewer's) life and are frequently drawn from real client experiences."

And on the layout, which is a teaching choice rather than a design one: "you may have noticed that you are reading this compilation in landscape format. This is intentional. Consultants think in terms of PowerPoint slides much more often than essay-style documents."

The warning nobody passes on

Read the right column against how the file is actually used. A candidate working through write-ups to learn the correct answer to each prompt is doing something the authors explicitly warned against, and the warning is in the first two hundred words of the document.

Two independent sources say the same thing from opposite sides of the table. The Ross board writes that two candidates can have very different conversations about one problem and that it encourages this. Bain publishes, on its own hiring page, that "we're not looking for a single right answer; we're looking for how you think, structure problems, and build on ideas as the conversation evolves."

A club and a firm, with no relationship to each other, describing the same fact. That convergence is the strongest thing on this page, and it is what a model answer in any case book has to be read against.

There is a second convergence worth naming. The Ross board says its problems are "frequently drawn from real client experiences". Accenture's own case workbook says its prompts "tend to be real business situations, often drawn from the interviewer's actual project experience". The provenance claim matches. What differs is who remembered it, and how long ago.

Who could own the Ross case book, and why there is no download

The document has at least three possible rights holders and none of them has published a policy.

The club that issued it. The thirteen named case writers, each credited with an individual case. And the consulting firms whose interviews were reconstructed, since the board says the cases record real interviews that members sat.

So this page contains no download, no link to a copy, and no reproduction of any prompt, exhibit, dataset, solution or case list from the document. What we quote is the letter the board wrote about the document, which is a different thing from the document.

The copy we read is itself part of the pattern. It sits on the career services CDN of another university, which republished a Ross club's work without any visible permission. That is the same behavior as the aggregators, performed by an institution.

For a Ross student the route is the club. For everybody else, the firms publish prompts whose provenance is a live engagement rather than a memory, and the last sections here are what to do with one.

Three things called a casebook at this university

The results for this query are polluted in a way no other school in this batch is, because the word has a third meaning here.

The university's own institutional repository holds a record titled "Ross School of Business - Instructional track casebooks", dated 2018 and deposited by the university's human resources office. Both its files are marked restricted to on-site access at a historical library. Those are administrative documents and have nothing to do with consulting interviews. We opened the record specifically to check, and it ranks on this query.

The second meaning is the club document above. The third is other schools' casebooks republished on career service CDNs, which is how we ended up reading a 2013 Michigan-attributed copy hosted by a different university.

One more artifact deserves a mention, because it shows how far these files travel. A recording titled "Self-Checkout, Ross 2010" was made by the Yale SOM Consulting Club and is catalogued as freely available by Kellogg's career library. A Ross prompt, worked by Yale students, listed by Northwestern. We work all four of those recordings on our Yale case book page.

The club URL that stopped being the club's

There used to be a Ross club page indexed under a heading promising casebooks and other resources. On 1 October 2026 that address no longer serves anything from the school. It redirects to the commercial site of the company that makes the student group platform, where the content is product documentation. The school's subdomain on that platform returns 403 to the same vendor site.

This is the second club in this batch whose infrastructure has gone. The Darden club's own domain stopped resolving entirely, which we cover on our Darden case book page.

Draw the general lesson rather than the specific one. A club board turns over every twelve months and its hosting turns over with it. An indexed URL bearing a club's name is not evidence that the club publishes anything today, and a search result is not a check. The check is a request, and the request takes one command.

That also explains why these documents circulate as detached files. The institution that made them does not maintain an address for them, so the copies outlive the publisher.

What Ross publishes about consulting, in four separate lines

Ross splits the consulting function four ways, which no other school in this batch does. From its Class of 2025 employment data, read 30 September 2026.

The four components sum exactly: 28.4 plus 5.1 plus 1.7 plus 1.7 is 36.9. The function total exceeds the industry figure by 3.4 points.

The medians are the finding. Management consulting pays $190,000 and technology consulting pays $130,000, a $60,000 gap inside a single word at a single school in a single class. Yale publishes two consulting lines and Columbia publishes two; Ross publishes four, and the spread it reveals is the largest of the three.

One comparability warning. Ross reports that 92 percent of graduates received a first offer within six months of graduation. Six of the eleven schools in this batch report at three months, so that figure is not comparable as printed, in the same way Kellogg's is not, which we set out on our Kellogg recruiting page. The class median base was $170,000 and the median bonus $30,000.

Working a Ross prompt the way its board describes

The board's warning is also an instruction. If two candidates can have very different conversations about the same problem, then the write-up is one of those conversations and yours is another.

1. The partner reads the situation and the question once, then stops. No exhibit, no hint, no additional sentence, whatever the pause costs.
2. Ninety seconds on paper, announced aloud. The partner calls the start and the stop and holds to both.
3. Deliver the structure aloud and defend the shape. Name each branch and say why it fits this business rather than any business of the type.
4. Data on named request only. The partner releases what you ask for and nothing else, and you say the unit before every product.
5. One sentence of recommendation, answer first. Then the supporting number, then the risk that would reverse it.

Then swap seats. Accenture publishes the only duration any firm gives, "case interviews typically last from 45 to 60 minutes, with a focus on the case itself lasting 30 to 45 minutes", so an hour covers a pair working one prompt each way.

Reading a case book write-up the board told you to argue with

Open the write-up after your recommendation is spoken. Read your partner's notes first, because those are about your performance and the write-up is about the problem.

Then use the board's own words as the frame. Three questions, one written line each. Which branch of their solution was absent from your structure. Where your arithmetic diverged from theirs, if it did. And whether they reached a different conclusion, or the same conclusion by a different route.

A divergence on the third question is the interesting outcome rather than a failure, and the club that wrote the document says so. What would be a failure is reaching their conclusion without being able to say why, which is what memorizing a write-up produces.

Where a write-up carries an exhibit, cover its commentary and read the chart yourself first. Chart reading is scored on its own, and a write-up hands you the interpretation free, which is the one thing an interviewer never does. The landscape slide format the board chose makes that harder, because commentary and chart sit side by side, so cover it with a hand.

Naming the dimension after a Ross practice case

Write a miss down only if it names something that can happen again. A lost cost branch is a coverage failure, and it will reappear on an unrelated industry inside a fortnight. A single forgotten line item inside a single prompt carries no information and belongs in no log.

The four numbers below come out of src/lib/grading/rubrics.ts in our own repository. They are ours, and no consulting firm publishes a comparable scheme, which is why this sentence sits above the table.

Nearly half the structuring score sits on Coverage, and the reason is the shape of the failure. An answer that loses a whole branch stays lost however precise the surviving branches are. Translating the protocol above into those four names is mechanical. The branch a partner recorded as absent belongs to Coverage. Two branches counting one dollar twice belong to Separation. A structure delivered inside the clock with nothing specific under any branch belongs to Depth.

The six rubrics are laid out one by one on how an answer is assessed, each with the argument for its weights. The structuring drill and everything else in the tools runs without an account, on prompts we wrote ourselves, and returns a named first correction rather than a number. Three limits go with that: nothing in it acts as an interviewer, nothing in it teaches business content, and our six assessment simulators are all unshipped, as the assessments page states.

What changed on the Ross page, and the pointer deleted

1 October 2026. Rewritten after the 2019 board letter was read in full, after the university repository record was checked and set aside, and after the school's Class of 2025 employment data was read through a text proxy. The earlier version ran 269 words and described what a case book is good and bad for without quoting the one document in this whole subject whose authors explain themselves.

The old text finished by sending readers to our free resources page for case books from several schools. That promise was false when it was written. On 29 September 2026 a check of this repository turned up no PDF, no ZIP and no archive of any kind, and the three-way rights problem described above is why none will be added later. Eleven pages carried that pointer and all eleven have lost it.

Three controls stay red on purpose. We hold no individual account from a Ross student who worked this document; the board letter is a collective statement by seventeen named people and we label it as one. We have no transcript of the Yale recording of a Ross prompt, so nothing spoken inside it is quoted. And we cannot say who holds the rights, because three categories of possible holder have published nothing between them. Every attempt and its date is in the research file.