Case Interview Checklist and Rubric Prep

Most candidates do not need another broad explanation of case scoring. They need a way to turn one mock into better practice the next day. This page gives you a simple before, during, and after checklist so you can use every rep to improve. It complements the broader case interview scoring rubric without repeating it, and it works well with the case interview prep guide, consulting interview prep timeline, frameworks guide, math practice, synthesis guide, case interview communication tips, and behavioral interview consulting pages.

What Problem This Checklist Solves?

The common failure in case prep is false progress. You do a mock, get a few comments, nod along, and move on. That feels productive, but it does not produce clean learning. A checklist forces specificity. Instead of "need to be more structured," you get "my opening branch had overlap" or "I did not justify why I chose a profitability tree."

That matters because case performance is built from habits, not awareness. If you cannot name the exact mistake, you will repeat it in the next mock. A checklist gives you a shared language for what went wrong and what to practice next.

The right mindset is simple: each mock should produce one or two concrete actions. Not ten. Not a rewrite of your entire prep plan. Just the next rep.

How Should You Prepare Before the Mock?

Set one goal for the case

Pick one primary focus before you start. Examples:

Execution checklist

• ✓

  Keep my opening structure MECE. A MECE opening keeps every later branch anchored to a real driver instead of overlapping categories that blur the analysis.

• ✓

  Lead every calculation with the setup. Naming the formula before you compute proves you understand the logic, not just the arithmetic.

• ✓

  Give a one-sentence synthesis at the end of each branch. A branch-level synthesis keeps the interviewer oriented and shows you are tracking the story, not just crunching numbers.

• ✓

  Pause before answering data questions. A beat of silence before you answer signals you are thinking, not guessing, and catches misreads before they compound.

If you try to improve everything in the same case, you will improve nothing cleanly. One focus makes the review usable.

Pick the rubric lens you want to test

Do not wait until the end to decide how you are measuring the rep. Use a simple lens based on the interviewer rubric: structure, math, communication, judgment, and synthesis. That is enough. You do not need a giant form.

If you want the broader partner-style view, skim case interview scoring rubric first, then use this page to make that rubric actionable in practice.

Define the case conditions

Write down the case type, the format, and the time limit. A mock case feels different depending on whether you are doing a live interview, a solo drill, or a timed AI session. Your review only makes sense if you know the conditions.

For example, a weak opening in a 5-minute timed drill matters differently than a weak opening in a full 30-minute case. Context changes the diagnosis.

What Should You Watch During the Case?

Track the opening separately

Your opening is where you show whether you can own the problem. Check whether you:

Execution checklist

• ✓

  Restated the objective clearly. Restating the objective proves you are solving the client's actual question, not a version you assumed.

• ✓

  Asked clarifying questions that mattered. Clarifying questions that change your approach show judgment; questions that do not move the analysis just burn time.

• ✓

  Built a case-specific structure. A structure built for this client's situation, not a generic template, is what interviewers are actually scoring.

• ✓

  Explained why your branches fit the problem. Naming the logic behind each branch turns a list of buckets into a defensible argument.

If the opening is muddy, the rest of the case starts on a weak footing. That is why the frameworks guide matters, but this checklist is about whether you actually used a framework well in real time.

Watch your math process, not just the answer

Do not only mark whether the number was right. Mark whether you:

Execution checklist

• ✓

  Named the formula or setup. Stating the formula before you calculate lets the interviewer follow your logic and catch a wrong setup before it compounds into a wrong answer.

• ✓

  Kept units straight. Mixed units are the single most common way a correct approach produces a wrong number.

• ✓

  Rounded intentionally. Deliberate rounding shows control over the math; sloppy rounding reads as guessing.

• ✓

  Sanity-checked the result. A number checked against a benchmark is a defended answer; a number with no check reads as a guess.

This distinction matters because a correct answer with a bad process is fragile. It will fail under pressure in a harder case. A clean process survives harder numbers.

Notice how you communicate transitions

Candidates often know the content but lose points when they move between ideas. During the case, notice whether you signpost:

• "I have two hypotheses."
• "There are three drivers here."
• "I want to test the volume side first."
• "My recommendation so far is..."

If you want a deeper view of communication and synthesis mechanics, cross-reference the synthesis guide and keep this checklist focused on execution.

Record any moments of drift

Drift is when you wander from the original objective. It can happen in a few ways:

• You explore a branch without saying why
• You chase a number that does not change the decision
• You answer the interviewer's prompt but not the case question

Mark those moments immediately. They are often more useful than the final score because they reveal decision errors, not just presentation errors.

How Do You Review the Case Afterward?

Score the rep in one pass

Use a simple 1 to 4 scale or a pass, partial, fail system. Keep it stable across sessions. The point is consistency. If you change the scale every time, you lose the pattern.

Write scores for:

• Opening structure
• Analytical accuracy
• Communication
• Hypothesis management
• Recommendation quality

Write the one-line diagnosis

After the score, write one sentence that starts with the real issue, not the symptom. For example:

• "My structure was generic, so the rest of the case had no focus."
• "My math was fine, but I never explained the business meaning."
• "My synthesis came too late and lacked a clear decision."

That sentence is the bridge between feedback and practice.

Turn diagnosis into a drill

Every miss should map to one drill. Examples:

• Weak opening structure -> do structure drills focused on opening frameworks.
• Weak math setup -> run timed reps from case interview math practice.
• Weak closing -> practice a 30-second recommendation from the synthesis guide.

If you want a wider practice loop, use free case practice to track your reps and diagnose where the gaps show up most often.

Which Mistakes Should You Flag Every Time?

Execution checklist

• ✓

  Generic framework use. Branch names that could fit any case are a red flag; you want case-specific language, not a memorized template.

• ✓

  Unexplained math. Jumping straight to a number hides your thinking; interviewers need the setup to trust the answer.

• ✓

  Weak synthesis. Ending with 'I think they should probably consider...' means the rep was incomplete, not finished.

• ✓

  No explicit tradeoff. Skipping the tradeoff or risk in your recommendation means you may be skipping the real business tension.

• ✓

  No follow-up note. If you finish the case and do not write down the lesson, the feedback evaporates.

How Should You Use This in a Weekly Prep Plan?

Use the checklist in a tight loop:

1. Pick one focus before the mock.
2. Run the case.
3. Score the rep right after.
4. Write one diagnosis sentence.
5. Choose one drill for the next session.

That is enough. You do not need a huge tracker to start. You need repetition plus a narrow review habit. A natural Road to Offer routine is one full mock case, one checklist review, and one targeted drill before the next mock.

The fastest improvement usually comes from the weakest repeated pattern, not from trying to optimize every part of the case at once. That is why this page pairs well with the broader rubric article, case interview prep guide, and consulting interview prep timeline. The checklist tells you what to notice. The drills tell you how to fix it.

How Do You Score a Case When the Feedback Disagrees?

Treat feedback as evidence, not a vote. Translate every comment into an observable behavior, compare it with the transcript or recording when one exists, and ask whether the same miss appears in comparable reps. A confident comment with no example should not automatically outweigh a quieter comment tied to a specific moment.

Resolve disagreement conditionally:

1. If two reviewers cite the same moment, combine their labels and diagnose the observable behavior.
2. If the labels differ but the evidence matches, keep the evidence and choose the narrower diagnosis.
3. If one comment conflicts with the transcript, downgrade it unless the reviewer identifies context the transcript missed.
4. If feedback comes from different case types, do not call it contradictory until you compare like with like.
5. If no source gives an example, mark the claim unconfirmed and design one rep that can test it.

Which Feedback Source Should You Trust: Peer, Coach, AI, or Transcript?

Use each source for what it can observe. A transcript is strong for exact wording and sequence but may omit delivery. A recording captures pacing and interruptions but needs a rubric. A peer or coach can judge the effect on a listener, but should still point to evidence. AI feedback can surface patterns in the material it receives, but its conclusion should be checked against the underlying response and product capabilities should be confirmed in the current flow.

The source hierarchy depends on the claim. For “I used the wrong denominator,” written math is primary. For “I sounded defensive,” a recording and specific listener observation are more relevant. For “my framework omitted capacity,” the structure itself settles the point.

What Should a Useful Practice Log Record Besides Case Count?

Record one row per comparable rep:

Do not log “bad structure” or “need confidence.” Those labels do not define the next action.

How Do You Identify a Stable Weakness From Noisy Mocks?

Group comments by observable behavior, then rank them using four questions:

1. Has the behavior appeared at least twice in comparable reps?
2. Did it change the decision or block progress?
3. Is the evidence direct and specific?
4. Can one narrow rep isolate it?

Prioritize repeated, decision-changing misses with direct evidence. A one-off style preference belongs below a recurring error that changes the answer.

Worked contradictory-feedback log

After three mocks, a candidate receives these notes:

• Peer A: “Your structure was too detailed.”
• Coach B: “You needed more detail in the market branch.”
• Transcript: the opening had seven sub-branches, but none tested whether the new product would cannibalize the existing one.
• AI review of the supplied transcript: “The structure names many factors but does not prioritize the decision mechanism.”

Resolution. Accept the transcript evidence. Downgrade the apparent contradiction because “too detailed” and “needed more detail” refer to different properties. The stable miss is not the amount of detail. It is unprioritized detail and a missing cannibalization test.

Next week's plan. Run three opening structures for adjacent-product launches. Limit each to three top-level branches. Require one explicit cannibalization branch and one sentence explaining which branch comes first. End the week with one full case to test whether the correction transfers.

Readiness decision. Not blocked from interviewing solely by these three notes. The candidate passes this dimension when two comparable structures identify cannibalization and prioritize the first evidence request without prompting.

How Do You Stop Sounding Scripted and Rehearse Without Memorizing?

Rehearse decisions and evidence, not sentences. Change the numbers, industry, or exhibit order while keeping the underlying skill. If your answer breaks when the interviewer takes a different path, you memorized the path rather than learned the method.

For behavioral practice without a partner, record one answer from five anchors: situation, decision, two actions, result, reflection. For a friend new to consulting, give them one observable check at a time, such as whether you stated the conclusion before the detail.

How Do You Know Whether You Are Ready to Interview?

There is no universal score that proves readiness. Use a gate based on recent comparable evidence:

• No repeated decision-breaking miss remains untested.
• You can recover from one error without losing the case objective.
• Your math and exhibits preserve units and implications.
• Your final recommendation includes decision, evidence, risk, and next step.
• Performance holds across more than one case type and more than one feedback source.

If a gate fails, prescribe the narrow rep. If all gates hold, additional full cases may add fatigue more than information.

How Do You Rebuild Confidence While Keeping the Reps Realistic?

Separate confidence from evidence. After a bad mock, run a smaller rep that preserves the failed skill, then retest it in a full case. Do not replace every difficult case with easy practice, and do not repeat full cases without diagnosis.

For live in-interview errors, use the case interview recovery playbook. This page owns the post-case feedback loop.

Sources and Further Reading (checked 2026-06-10)

• BCG case interview preparation: careers.bcg.com/case-interview-preparation
• McKinsey interviewing: mckinsey.com/careers/interviewing
• Bain case interview preparation: bain.com/careers/hiring-process/case-interview
• Princeton Case Interview Preparation: careerdevelopment.princeton.edu/guides/interviews/case-interview-preparation
• Case interview scoring rubric