Our honest promise

Honesty is not a nicety here. The people who judge a consulting interview detect vague or invented reasoning for a living, and the firms a candidate is aiming at would catch any exaggeration. So we are straight about exactly what this product is.

What it does

Makes you genuinely ready on the things that can be taught and practiced: structuring, math, exhibits, every case type, the firm formats, the digital screens, and the fit interview.
Gives realistic on-demand practice that would otherwise cost a lot from a human coach.
Builds a real problem solver who could take on actual client problems, which is what the interview screens for.
Teaches from original material, never copied from a firm or a paid case bank, with every calculation in the worked cases checked automatically.

What it does not do

Guarantee you an offer. No one can promise the outcome on the day.
Fully replace doing live practice cases with real people, which still matters.
Provide or reproduce any firm's actual proprietary test. We build the underlying skills and practice with similar formats.
Use manufactured difficulty, padding, empty praise, or scripted gotchas to run up a bill.

How we grade

Every score on this site comes from published rules, applied the same way to everyone. No score is a hidden judgement, and none comes from an AI. Here is every rule, and, just as important, what each one cannot judge.

Every number in every worked case is recomputed by a checking script before the site is published, and publishing stops if one is wrong.
The same answer always gets the same score. Your results show which checks passed and why.
An AI opinion, if one is ever offered, is labelled as one and never counts toward a level or the Project Fellows bar. Neither does a session you rated yourself: it shows in your trends, never in your level.

NumbersShow the rules

Every number you type in the case simulator, and the Mental Math Gym.

Correct: Within 1.5 percent of the worked answer, or the worked answer rounded the way people round in an interview. Full credit.
Close: Within 6 percent. Half credit, with a note to round only at the end.
Right digits, wrong size: Off by a power of ten (a thousand instead of a million). Half credit, with a note on units.
Right size, wrong direction: A fall written as a rise. Half credit. A fall written as a positive number with a word like "fell" or "down" counts as correct.
Reading what you typed: Million, billion, crore, lakh, percent signs, pence and cents, and accountants' brackets for a negative are all understood. If you type several numbers, only the last is graded, and you are told.
Mental Math Gym: Every reasonable reading of what you typed is tried against the exact answer, within the small tolerance each problem states.

What this cannot judge: Whether your method was sensible. A right number reached by luck scores the same as a right number reached well, so the worked steps are always shown.

StructuresShow the rules

A typed structure in the case simulator, and the Structure Builder.

Branch coverage: Each branch of the model structure counts as covered when your answer uses one of its own key words, or two of the words it shares with other branches. Credit is the share of branches covered.
Laid out: A structure written as one run-on sentence loses a tenth of the credit; one per line, or first, second, third, is enough.
Length: Over 180 words loses a tenth: detail comes later in the case.
Structure Builder: Your chosen buckets are compared with each valid split of the problem. Credit is the right buckets found minus the ones that do not belong, over the size of the split. It is MECE when you have exactly one full split.

What this cannot judge: Whether a different structure of your own is good. A structure can be sound and still use other words than ours, so always compare yours with the model.

The recommendation rubric (10 points)Show the rules

The closing recommendation in the case simulator.

Answer first: the first sentence gives your decision (or a clear conclusion): 2 points, with partial points where the rule allows.
Uses numbers from the case (at least two): 2 points, with partial points where the rule allows.
Gives reasons (because, due to, or sentences that open with First, Second): 2 points, with partial points where the rule allows.
Names a risk: 1 point.
Proposes a next step: 1 point.
Mentions the key ideas of this case (at least two): 1 point.
Concise: 30 to 175 words: 1 point.
Answer first: the first sentence states your answer (the number or the conclusion): 2 points, with partial points where the rule allows.
Names a risk or the key assumption: 1 point.
Answer first: the first sentence gives your answer (your decision or your conclusion): 2 points, with partial points where the rule allows.
Result: Eight points or more counts as a strong close.

What this cannot judge: Whether your decision is the right one. The rubric checks the shape of your answer, so compare it with the model recommendation.

Exhibit takeawaysShow the rules

Typed chart takeaways in the case simulator, and the Exhibit Gym.

Names the item that matters: 2 points of 4. The item is the one the model answer names, shown in brackets on your result.
Quotes a number from the exhibit: 1 point of 4.
Says what it means: 1 point of 4. The item is the one the model answer names, shown in brackets on your result.
Choosing an answer: Where you pick from options, one right answer earns the point. Where you pick several areas, credit is right picks minus wrong picks, over the number of right ones.

What this cannot judge: How clearly you explained it. The check looks for the item, a number and a linking word.

Market sizingShow the rules

The Market Sizing Gym.

Close: Within 1.5 times the model answer, either way.
Same ballpark: Within 3 times. This still counts as on track for your session score.
Off: More than 3 times away.

What this cannot judge: Whether your assumptions are reasonable. Two good estimates can differ, so the model answer shows its steps and sources for you to compare.

Online test practiceShow the rules

The practice sets in Online test practice.

Typed numbers: Whole-number answers must be exact to the nearest whole number. Decimal answers may be one step off in the last decimal.
Answer first: Passed or not, with the reason shown.
Two numbers from the data: Passed or not, with the reason shown.
One caveat: Passed or not, with the reason shown.
About four sentences: Passed or not, with the reason shown.
Constraint puzzle: Your pick either meets every condition or it does not; each condition is shown with your total against its limit.
Check the AI: Each paragraph marked keep, fix or drop against the planted problem it holds; corrected numbers within their stated tolerance; the missing key factor; and a client question that ends with a question mark and names that factor. Dropping a sound point is pointed out.

What this cannot judge: How you worked. Real tests may also score your process, which is not public, so these sets do not pretend to score it. They are practice in a similar format, not the real test.

Fit storiesShow the rules

The structure check in the story workshop.

Situation and task: Each is filled in with enough to give the listener context.
Action: At least 60 words on what you did, in order.
Result: A concrete result that includes a number.
Your own role: You say "I" at least as often as "we" in the action.
Length: 150 to 450 words, about one to three minutes spoken.

What this cannot judge: Whether the story is persuasive or true, or how it sounds out loud. Tell it to a partner and ask the follow-up questions.

CV and cover letter checkerShow the rules

Check my CV, on the Applications and CV page.

About one page: Counts the words. Up to 650 usually fits one page for a student or early-career CV; under 200 is probably unfinished.
Bullets open with an action verb: At least 70 percent of bullets should start with one of the 338 action verb forms on our list (led, built, ran and so on). British and American spellings both count, for example analysed and analyzed.
Bullets show a result with a number: At least 50 percent of bullets should hold a number (a digit, or doubled, tripled, halved).
No first person in bullets: Bullets should not use I, me, my, we, our or us. The reader already knows the CV is about you, and "we" hides your own role.
No duty phrases: Flags phrases that describe a duty instead of an action, such as "responsible for", "helped with", "worked on" and "involved in".
Shows qualities instead of listing them: Flags words nearly every applicant uses about themselves, such as "hard-working", "team player", "detail-oriented" and "passionate". A reader cannot check them, so a line that shows the quality with a result does more. On a CV any of them is flagged; in a cover letter, more than 2 different ones.
One date style: Finds dates written as "Jan 2025", "January 2025", "01/2025" or "2025-01" and flags a CV that mixes styles.
Tense is consistent: Notes bullets that open in the present tense ("Lead") alongside the past tense ("Led"). The present tense is right only for a role you still hold.
Bullets are short: Flags bullets over 30 words, about three printed lines.
Shows what consulting looks for: Looks for words that show leadership, analysis, results, teamwork and business sense. At least 4 of the 5 should appear somewhere.
Formatting is consistent: Flags mixed bullet symbols, bullets where some end with a full stop and others do not, and double spaces.
No common misspellings: Checks against a small list of words people often misspell (for example "managment" and "recieve"). British and American spellings are both fine. It is not a full spelling check.
Personal details fit the region: Looks for a date of birth, age, marital status or nationality, and compares with the general norm on these (and on photos) for the region you choose. Follow the firm when it says otherwise.
Names the right firm: The firm you typed appears in the letter, and no other firm from our list does.
Says why this firm, specifically: A sentence that mentions the firm (or "your" firm, office or team) together with something specific: a conversation, an event, a report, an office, a practice or a training scheme.
Under one page: Between 150 and 400 words, and within the word limit you set, if any.
Gives proof with numbers: At least 2 numbers, so your examples show a result.
Says why consulting: Mentions consulting or the consultant role at least once.

What this cannot judge: These checks look at form, not substance. They cannot tell whether your experience is strong, whether a number is true, whether the CV suits a particular firm, or how a recruiter will read it. They are not an AI and do not rewrite anything for you.

Speaking statsShow the rules

Spoken answers and the practice call room transcript, when you choose to use them.

Words per minute: Words in the transcript over the time you spoke.
Filler words per 100 words: Counted from a published list (for example "um", "umm", "uhm", "like", "you know"); some words count only when set off by commas. Speech recognition drops some hesitations, so this is a lower bound.
Length against the time limit: How long you spoke compared with the time the question allows.

What this cannot judge: Tone, confidence or emotion. Those are never scored from your voice or your camera; the stats are plain counts.

The candidate standards, version 2

Your scorecard places each skill at one of 4 levels (Building, Interview-ready, Top-firm ready, Fellow-ready) from your graded practice in the last 90 days. The thresholds were set by judgement and will be checked against the interview results members report. A level describes practice on this site; it does not predict an offer.

Change log

Version 1. The first published bar: seven skills, four levels, evidence from the last 90 days, and the Project Fellows bar.
Version 2. Thresholds unchanged. Sessions done elsewhere and rated by another person count as live evidence, like Partner mode; self-rated sessions never count toward a level; online test practice counts as graded practice for math, exhibits and synthesis only.

What no automated check can judge: whether your reasoning is sound, whether a structure of your own is better than ours, how you come across in the room, and whether you will get an offer. That is why live practice with another person still matters, and why every result shows the model answer next to yours.

The soul of it

An interviewer sees the same memorized framework ten times a day. The candidate who stands out actually understands the problem, narrows it to what truly matters, and reasons cleanly to the answer. We teach the frameworks that come up in case interviews, but always as a tool for thinking, never as a script to recite. The real test of our success is not only that you pass the interview, but that you could sit down in your first months at a top firm and take on real client problems.