I keep getting the same generic feedback after mocks: “be more structured,” “tighten your math,” “clearer synthesis.” Helpful, but not specific enough to fix quickly. I’m trying to move to an actual interviewer-style scorecard so I can track progress by bucket rather than vibes.
If you’ve run real interviews, what did your rubric actually measure and how did you anchor the scores? I’m thinking structure, math, synthesis, and communication as the core, but I don’t know where people draw the line on “meets bar” versus “strong” in each. For example, what does a solid opening actually include in your eyes, and how fast is fast enough on math without sloppiness? Same for synthesis: what’s the minimum viable recommendation that still sounds like a recommendation, not a recap?
If anyone’s open to it, I’d love to run a 25–30 minute mock this week and get a score per bucket with a few lines of written feedback. Even a redacted template would help. What does your real scorecard look like, and would you share a blank copy or sample comments?
here’s the boring truth: most of us used four buckets—structure, quant, synthesis, comms—on a 1–4 scale. 3 means “hireable today,” not “case god.” anchors were simple: coherent tree without fluff, math set up clean with a quick sense check, one clear answer with 2 risks and a next step, plus crisp signposting. if you’re haggling over 2.7 vs 3.1 you’re missing the plot. keep it readable, not academic. and yes, we really do ding rambling.
don’t overengineer the timing either. opening that runs past 90 seconds smells like filibuster. math: one minor slip is fine if you catch it, unit confusion is an auto-2. synthesis needs a headline in the first sentence and an action in the last—no storytime. i’ve used a simple rule: if i can’t jot your recommendation in one line, you didn’t synthesize, you narrated. harsh, i know, but saves candidates months of wheel-spinning.
i’m hunting for this too. happy to swap a 30‑min mock this week and plug scores into a sheet. dm me if you’re down, we can use a simple 1–4 per bucket. ty!
When I trained interviewers, we defaulted to a four-bucket rubric with anchored behaviors. Structure evaluates whether the candidate frames the objective, states a directional hypothesis, and lays out mutually exclusive drivers without spraying sub-branches. Quant looks at setup correctness, unit hygiene, and whether the candidate tests reasonableness unprompted. Insight is about identifying the decisive driver and implications for the recommendation, not just listing observations. Synthesis requires a one-line answer, two driver-based supports, explicit risks, and a concrete next step. Weights were roughly balanced, but a weak synthesis dragged the whole score. “Meets bar” is consistently clear, not flashy. If you share notes, I’m happy to mark them against that standard and write a few example lines so you can calibrate.
Love this ask! Clear rubrics change the game. If you post your notes, I’ll help score one mock this weekend. You’re closer than you think—keep going!
I flailed until a former Bain buddy sent me his old score sheet. The thing that helped most was the “green/yellow/red” comments next to each bucket. He wrote stuff like “framework clean but no hypothesis, yellow” and “math setup perfect, green.” Seeing that side-by-side made it obvious where I was leaking points. After two sessions using that sheet, my openings got under a minute and my synth finally sounded like a decision, not a summary. Happy to share a cleaned version if you want.
A practical scoring template that tracked well in my cohorts used five dimensions with weights: Opening/Objective (15%), Structure (25%), Quant (25%), Insight/Business Judgment (15%), Synthesis/Communication (20%). Time benchmarks: opening ≤60–75s, math setup ≤30s before computing, final synthesis ≤60s. Anchors: a 3 in Structure requires MECE-enough branches tied to the stated objective and a directional hypothesis; a 3 in Quant requires correct setup, units, and self-initiated sanity checks. We flagged two recurring failure modes: wandering openings and unit drift. If you post a mock write-up, I can score it against this grid.