What to take away
- Generic feedback like 'be more structured' fails because it names a category of skill, not a moment you can actually go back and change
- Useful feedback has three parts: the exact moment, what a stronger version would have sounded like right there, and which one skill to drill next
- A peer with no interviewing background can still give you useful feedback if you hand them three or four specific things to watch for before the case starts
- Recording your own practice sessions turns vague self-assessment into evidence you can check yourself against instead of a feeling you trust or don't
- One specific note acted on beats five vague ones collected and never opened again
- The right question after a case is not 'how did I do' but 'what's the one thing I should do differently next time'
Why does most case interview feedback sound the same
Ask around after a mock case and you'll hear the same five or six lines on repeat: be more structured, work on your math, slow down, be more confident, practice more. Almost every candidate we've talked to has heard some version of all of them, often after completely different cases, from completely different partners.
That repetition isn't a coincidence. It's what happens when feedback gets generated under three pressures at once: the person giving it doesn't want to be harsh, doesn't have a framework to pin the comment to, and is trying to summarise thirty minutes of conversation from memory a few seconds after it ended. "Be more structured" survives all three pressures. It's true often enough to feel safe, vague enough to feel kind, and short enough to say without notes.
None of that makes it useful. A category isn't an instruction. If someone tells a writer "be more concise" without pointing at the paragraph that rambles, the writer has been handed a goal, not a way to reach it. Case feedback has the identical problem, just less obvious, because "structure" sounds specific even when it isn't attached to anything.
There's a second, quieter reason generic feedback survives: giving specific feedback is genuinely uncomfortable in a way giving a category isn't. Telling a friend "be more structured" costs nothing socially. Telling them "you said 'I think, maybe' before three different numbers in the last ten minutes, and it made me doubt figures you'd actually gotten right" requires you to have been paying that close attention, and to say something that could sting. Most practice partners, understandably, choose the version that doesn't require either.
What's the actual difference between generic and specific feedback
Specific feedback and generic feedback can be about the exact same weakness and still be worlds apart in usefulness. The difference is never the topic. It's whether the feedback points at a moment.
| Generic version | Specific version |
|---|---|
| Be more structured | When you got the profitability question, you jumped straight into revenue drivers without naming your buckets first. Say the three branches out loud before you go deep into the first one |
| Work on your math | You got the market size right but didn't sense-check it against the population number you'd already used two minutes earlier. State the sanity check out loud, even a rough one |
| Be more confident | You said "I think, maybe" before three separate numbers in the last ten minutes. State the number, then the caveat after, not before |
| Practice more | Your synthesis at the end didn't include a next step. Add one sentence after the recommendation: what would you actually do Monday morning |
Notice what the right-hand column has that the left doesn't: a moment ("when you got the profitability question"), and a specific alternative ("say the three branches out loud first"). Both are doing real work. The moment tells you where to look on your next recording. The alternative tells you what to try instead, rather than just what was wrong.
The one-sentence version of the difference
Generic feedback describes a trait. Specific feedback describes an edit. "You're not structured" is a trait, and traits feel permanent, which is part of why they're demoralising to hear. "Name your buckets before you go deep into the first one" is an edit, and edits are things you can just do differently next time.
Why doesn't "be more structured" actually help anyone improve
It's worth taking this one line apart specifically, because it's the single most common piece of case feedback in existence and it fails in a very particular way. Structure isn't one skill. On our own rubric, what gets called "structure" in casual feedback usually maps to framing, and sometimes bleeds into commercial judgement (picking the right branch) or synthesis (structuring the close). Telling someone to "be more structured" doesn't say which of those it is.
A candidate who hears it has three or four plausible fixes available and no way to know which one applies to them. Build a cleaner framework upfront? Slow down before diving into a branch? Signpost more while talking? Close with a clearer recommendation? Most candidates guess, apply the wrong fix, hear the same note again next week, and start to believe structure is just a weakness they have rather than a specific, fixable habit.
- Ask which part: was it the opening framework, the choice of branch, or the closing recommendation
- Ask for the moment: which question in the case triggered the note
- Ask what a stronger version would have sounded like, in the interviewer's own words if possible
- Ask whether it happened once or repeatedly across the case
How do you get useful feedback from a case partner who isn't a trained interviewer
Most practice partners are exactly where you are: another candidate, prepping for the same interviews, with no formal training in what to listen for. That's not a reason to skip peer practice. It's a reason to structure it so a non-expert can still catch something real.
The fix is to hand your partner a short, specific watch-list before the case starts, rather than asking them to freestyle an evaluation from memory afterward. A partner who doesn't know what "commercial judgement" means can still tell you, accurately, whether you named your assumptions out loud or whether you talked over them when they tried to correct a number.
Give them two or three things to watch, not the whole rubric
"Notice whether I name my buckets before I go deep on the first one" and "notice whether I state a recommendation clearly at the end" are things anyone can track. A partner handed all fourteen rubric skills at once will track none of them well.
Ask them to write down the moment, not the verdict
"Around minute twelve, when you got the cost breakdown" is more useful to your partner and to you than "your math was okay." A timestamp or a rough marker beats a summary every time.
Ask what they would have said instead, even if they're guessing
A non-expert's guess at a better phrasing is still more useful than a vague verdict, because it gives you a concrete alternative to try, even one you end up improving on yourself.
Record the session so the feedback isn't the only record of it
Memory fades fast, including your partner's. A recording lets you go back to the actual moment they flagged and judge it yourself, rather than relying entirely on their after-the-fact account of it.
“Weak: "You did fine, just work on structure a bit."”
“Better: "At the start, when you got the market entry question, you went straight into 'well, first I'd look at market size' without naming your other buckets. I think you had three branches in your head, but I only heard one."”
“Weak: "Your math felt slow."”
“Better: "When you calculated the breakeven, there was maybe twenty seconds of silence before you said 'so roughly two years.' I couldn't tell if you were doing the math or stuck. Talking through it, even roughly, would've helped me follow."”
What should you expect from a coach, human or AI
A paid coach, human or AI, should clear a higher bar than a peer, because that's the actual value they're offering. If a coach's feedback still comes back as "good energy, work on structure," you're paying for reassurance, not for something you can act on.
The honest comparison between a human coach and an AI one isn't which is better in the abstract. It's what each is actually good at, and both have real limits worth knowing before you rely on either.
| A human coach | An AI coach |
|---|---|
| Reads tone, hesitation and body language in real time | Never gets tired of a candidate's fourth mediocre synthesis of the week and scores it the same as the first |
| Draws on years of pattern-matching across hundreds of real interviews | Can quote the exact sentence in your transcript that earned or lost a point, every time, without misremembering it |
| Can soften or escalate a note based on how a specific candidate is likely to receive it | Applies the identical written standard to every candidate, with no social pressure to go easy |
Neither column is a reason to pick one over the other outright. A human coach's pattern-matching is worth a lot and an AI coach can't replicate the judgement that comes from having sat across from five hundred real candidates. An AI coach's consistency is worth a lot too, especially for the unglamorous parts: a tired human giving you your third mock case of the day is more likely to default to "be more structured" than a system marking against the same written rubric every single time.
MBB Ready's coaches are AI, and we say that plainly rather than letting the name or the voice imply otherwise. What we ask them to produce is a skill-by-skill score against a published rubric, with the transcript quote that earned each one and a single thing to fix next, not a warm paragraph about how the case felt. Whatever coach you use, human or AI, ask for that shape and be wary of anything that arrives without it.
What does one complete, usable piece of case interview feedback actually contain
It helps to see the finished shape once, in full, rather than as a rule. A complete note has four parts: the skill it's about, a score or a clear read on where you stand, the specific evidence behind that read, and one sentence on what to do differently. Strip out any of the four and you're back to something closer to "be more structured."
Notice that neither note above needed a trained interviewer to write it, once someone had a specific skill in mind and a moment to point at. That's the whole method this guide is arguing for: not a better source of feedback, but a better shape for it, one that a peer, a coach, or your own recording can all be pushed to produce.
How do you give yourself useful feedback when you practice alone
Solo practice has one structural problem no method fully removes: you can't hear your own case the way an outside listener does, because you already know what you meant. Recording it and reviewing it as a stranger's transcript is the closest available fix.
- Record the audio, always, not just the notes you take afterward
- Wait at least a few hours before reviewing, ideally the next day, so you've lost the memory of what you meant and are stuck with what you actually said
- Write down three moments where you'd flag yourself if you were an interviewer, with a timestamp for each
- For each moment, write one sentence for what you'd say instead
- Pick one, not three, to actually practice before your next session
The step people skip is the wait. Reviewing a recording immediately after finishing a case means you're still inside your own intent, which is exactly the bias you're trying to remove. A day of distance turns "I meant to signal that I had a broader structure" into "I hear one branch and then silence," which is what an interviewer actually heard too.
What questions should you actually ask right after a practice case
The single biggest lever over feedback quality is the question you ask, not the person you ask it of. "How did I do?" invites a summary, and summaries default to the generic five lines. A sharper question forces a sharper answer, even from someone with no training.
| Weak question | Sharper version |
|---|---|
| How did I do? | What's the one thing you'd change if I ran this exact case again tomorrow? |
| Was my structure okay? | Was there a moment you couldn't tell what my overall structure was, even if I had one in my head? |
| How was my math? | Was there a point where you weren't sure if I was calculating or stuck? |
| Did I seem confident? | Was there a specific sentence where I hedged something I actually knew? |
Every sharper version on the right asks for a moment, not a rating. That single change, moment instead of rating, is most of what separates feedback you can use from feedback you can only nod at.
This works even mid-case, not just at the end. If you sense you're losing your partner's attention or your own thread, a quick "can I check something, did my last answer actually address what you asked" gets you a specific, immediate correction instead of a vague verdict twenty minutes later when neither of you remembers the exact moment clearly.
How much feedback should you actually be collecting, and how often
More feedback is not automatically better, past a fairly low point. Collecting notes from five different partners on five different cases in one week produces a pile of observations you'll never act on, because acting on feedback takes longer than receiving it. Receiving a note takes ten seconds. Actually changing the underlying habit takes several practice reps, deliberately aimed at that one thing.
A workable cadence: one focused piece of feedback per practice session, tracked against the same skill until it moves, then a new focus. If you're using a written rubric, that means picking your lowest-scoring skill after a session, drilling specifically toward it for the next two or three sessions, and only then checking whether it actually improved rather than just whether it feels different.
The sign you're doing this right is boring: the same skill keeps coming up as your focus for a stretch of sessions, because you're actually working on it rather than collecting a new note every time and starting over. The sign you're doing it wrong is a long list of different notes across different sessions that never gets shorter.
It's also worth naming the failure mode on the other end: candidates who take exactly one piece of feedback and stop practicing entirely once it's been "fixed" in a single session. A habit that took months to build rarely disappears after one good rep. Expect to see the same note recur, weaker each time, for a few sessions after you think you've corrected it, and treat that as normal rather than as evidence the fix didn't work.
Common questions
Why is 'be more structured' bad case interview feedback?
Because it names a category of skill rather than a moment. Structure can mean the opening framework, the choice of which branch to explore, or the clarity of the closing recommendation, and 'be more structured' doesn't say which. A candidate who hears it usually guesses at a fix, applies the wrong one, and hears the same note again next time.
How do I get better feedback from a friend who isn't an ex-consultant?
Give them two or three specific things to watch before the case starts, rather than asking for a general evaluation afterward. Someone with no case-interview background can still reliably notice whether you named your assumptions out loud, whether you talked over a correction, or whether your ending had a clear recommendation in it, if you tell them to listen for those things specifically.
Is AI feedback on a case interview actually useful, or is a human coach better?
Both have real strengths worth knowing. A human coach reads tone and hesitation and draws on pattern-matching across many real interviews; an AI coach applies the same written standard every time and can quote the exact transcript line behind a score, without fatigue or social softening. The useful question isn't which is better in general, it's whether whichever one you're using can point to a specific moment behind every note.
How specific should case interview feedback actually be?
Specific enough that you could act on it tomorrow, on a different case, without asking a follow-up question. That usually means three parts: the exact moment it happened, what a stronger version would have sounded like right there, and which one skill to focus on next. If a piece of feedback is missing all three, it's not finished yet.
How many practice cases should I get feedback on before a real interview?
Fewer sessions with feedback you actually act on beats more sessions with feedback you collect and file away. A workable rhythm is one clear focus per session, carried across two or three sessions until it visibly improves, rather than a new note every time. Somewhere before the real interview, run at least one full session under real time pressure with someone else asking the questions.
What if my feedback keeps saying the same generic thing every time?
Change the question you're asking, not just the person you're asking. 'How did I do?' invites a summary; 'what's the one thing you'd change if I ran this exact case again tomorrow?' forces a specific answer, even from a partner with no training. If the generic note persists after that, ask for the exact moment it happened before you accept the note as true.