Everything, end to end.

The green pill counts exhibits, the blue pill counts questions — a quick read on how quant- or discussion-heavy a case is before you commit. Common mistake: judging a case by duration alone; a short case with two exhibits is usually denser than a long discussion case.

A small blue count on a filter shows how many options you have active. Clearing search + filters brings the full library back.

Starting a case switches you to full screen for a real-interview feel — press Esc anytime to exit. If a case you built still has unresolved review flags, Start opens a review gate instead: clear the flags in the Case Builder first (General Principles, chapter 05).

Hover any control for its label. In Exam Mode the pause and hint controls are replaced by a lock icon — that's the whole point of exam pressure. Common mistake: muting yourself to think and forgetting to unmute — the screen tells you when your interviewer cannot hear you.

Pro tip: say the headline of the exhibit out loud before drilling in ("Revenue is flat while costs rise ~8% a year"). Describing before interpreting is itself a scored behavior — and it buys you thinking time.

Asking for a hint isn't penalized in your score the way a wrong answer is — but a real interview has none, so lean on them in Practice and wean off for Exam. Common mistake: sitting in frozen silence for minutes instead of taking rung 1 and keeping the case moving.

Use it the moment you start a calculation, not after she has already interrupted you — the point is to prevent the interruption, not recover from one. It is available in Exam Mode too: holding the floor is not help with the case, it just stops a pause being misread. Common mistake: going silent and hoping she waits. She cannot tell thinking from finishing unless you tell her.

Blocked the mic by accident? Click the camera/mic icon in your browser's address bar, allow the microphone, and start the case again. If a glitch genuinely wrecked a run, report it in the feedback modal (chapter 8) — reimbursement exists for exactly that.

The 100-word minimum is deliberate: "audio broke" is not enough to diagnose or reimburse. Say what happened, when in the case, and what you tried. If we reimburse you, a case credit is added and you'll see a thank-you banner in the Case Library next time you visit.

Scroll the debrief for the section-by-section breakdown and your prioritized improvements — the fastest way to know what to drill next. Common mistake: reading only the overall number; two 78s can hide completely different weaknesses.

Adaptability and Collaboration are only scored when they actually came up — a pushback you handled, a hint you used. No pushback, no penalty. Speed is measured on scored thinking time, not on how fast you talk.

Treat it like intervals, not a marathon: three reps back-to-back on the same case type teach you more than one "perfect" one. Volume is the feature — a full case gives you one opening a session; the Gym gives you ten an hour.

Hard isn't just longer — it's noisier. If your structures hold on Medium but wobble on Hard, the gap is prioritization, not frameworks: practice finding the question before answering it.

Before you speak, settle in: paper down, pen up, case type written at the top of the page. That quick ritual gets you set — and it's the same one that will steady you in the real interview.

Don't fish — interrogate. "Anything on costs?" is a fishing trip; "how do fixed and variable costs split?" is a question. Precision is what unlocks data here, and what impresses interviewers everywhere.

Your stars curate the Gym for everyone: 5★ problems circulate to other candidates, weak ones retire — and since the bank never deals you the same prompt twice, every rep stays fresh.

Notice the scores climbing bottom-to-top? That's the point. If a case felt terrible, still open the debrief — the worst sessions usually carry the most specific improvements.

Pro tip: trim the PDF before uploading — delete tables of contents, directories, and industry primers. You get more real cases per extraction, and the engine spends its effort on pages that matter.

Common mistake: writing the Key Insight into the spoken prompt — the candidate is supposed to reach it, not hear it. Keep the prompt to the question; keep the insight in its own field.

Opening a flagged chunk shows exactly what to check — often it is one figure against one quoted line of the source PDF. If you press Start on a case that still has blocking flags, a review gate opens instead of the interview; clear the flags and Start works normally.

"Pending" means someone joined through your link but hasn't subscribed yet — the reward lands automatically the moment they do.

The LinkedIn credit is one-time and one per account — the URL is verified as a real LinkedIn profile when you save it. If you saw the offer in the Case Library and dismissed it, this is where to claim it later.

Quota renews with your billing period; credits do not expire with it. If you are near your quota, spend credits — they were earned, not rented.