The written case is not a verbal case with more paper. In a live case the interviewer rations exhibits; here you get the whole pack at once and the clock is the interviewer. Most candidates lose before they write a slide, by reading every page as if the score sheet rewarded thoroughness.
The thesis: partners already know what is in the pack. They score whether you can throw most of it away, put a recommendation on slide one with the two numbers that carry it, and defend the rounding when they pull one assumption. Reading speed is not the skill. Triage is.
By the end you will have a time budget for a 60- or 90-minute invite, a reading protocol for a 20–30-page pack, a three-slide skeleton, and one fully worked example from prompt to slide text.
What the firms actually publish (and what they do not)
BCG's case interview preparation page describes the live case: structure the problem, ask questions, analyze data, do quick math, and communicate clearly. It does not publish a global written-case script. Bain's interviewing page says consulting roles typically include a digital assessment and case interviews, and that some roles include a written assignment you may later present. Bain's published FashionCo practice case is interactive, not a take-home pack, but it still grades the close the way a written deck should: a good answer names the option and a dollar gap; a better answer adds risk and next steps.
Big Four strategy arms sometimes put data in your hands without a full hour alone. Strategy& UK states on its careers FAQs that the structured first-round case gives you a slide pack reflecting real client work and about ten minutes to review it before recommending solutions. That is a cousin of the written case, not the same exam.
Office-level timing, pack length, calculator rules, and whether you write on blank slides or fill templates are candidate-reported and invitation-specific. Treat your recruiter email as the source of truth. The method below assumes a later-round written case of the kind candidates commonly describe at BCG and Bain: a data pack, limited solo time, a few slides, then partner Q&A.
Time budget for a 60- or 90-minute pack
Lock the clock before you open page two. Stretch reading and you will present a summary. Steal time from slides and you will present a pile of notes.
| Phase | 60-minute invite | 90-minute invite | What "done" looks like |
|---|---|---|---|
| Prompt + map | 5 min | 8 min | Decision verb written; exhibit titles logged |
| Priority read | 12 min | 18 min | Two or three exhibits annotated with one so-what each |
| Arithmetic | 15 min | 22 min | Two numbers that flip or support the call, with units |
| Slides | 18 min | 28 min | Three slides; exec summary states recommendation + two numbers |
| Rehearse defense | 10 min | 14 min | Weakest assumption named; rounding path written once |
If your invitation is closer to two hours (a length some BCG candidates report), keep the same proportions: roughly one-fifth mapping and reading, one-quarter math, one-third slides, remainder for defense notes. Do not invent a fifth calculation because you have time left.
Reading protocol for a 20–30-page pack
You will not finish the pack. That is the design. The protocol exists so you leave deliberate blanks instead of accidental ones.
1. Read the prompt twice (2–3 minutes). Circle the decision verb: enter, acquire, close, price, invest, divest. Write the deliverable in one line at the top of your notes ("recommend go/no-go on branch closure; 3 slides"). If the prompt lists questions, number them; your slides must answer those questions, not the ones the charts beg you to answer.
2. Build an exhibit log, titles only (3–5 minutes). Flip every page. For each exhibit write: page number, title, type (P&L, customer mix, competitor, ops, survey, appendix text). Do not read cells yet. After this pass you should be able to say which three pages could change the recommendation.
3. Rank, then read deep (10–15 minutes on a 60-minute clock). High priority: anything that moves revenue, cost, or capital for the decision. Medium: competitive or customer context that supports the mechanism. Low: industry color, long narrative, duplicate charts. Read high-priority exhibits carefully; annotate the one number you will cite. Skim medium for contradictions. Skip low unless a high-priority number points back to them.
4. One-line so-what per priority exhibit. "Corporate retention down 6 pts YoY; acquisition flat" beats a paragraph of chart description. If you cannot write the so-what, you are not done reading that page.
For how to pull insight off a chart without restating the axes, see case interview exhibits. For turning findings into a decision sentence, see case interview synthesis. For the live-case formats that sit beside the written pack, see the BCG case interview guide and the Bain case interview guide.
The arithmetic that actually matters
Partners probe calculations you put on paper. Three complete chains beat eight half-started ones.
Do only math that changes the recommendation: contribution or margin by segment, a gap versus target, a payback or break-even on the option you recommend, or a simple sensitivity (plus or minus 20% on the swing assumption). State formula, plug numbers, write units, then one implication sentence.
Skip vanity math: market size when the decision is already inside the market, CAGR on five years when the case turns on one year of mix, and NPV with a made-up discount rate when payback in years would do.
Ned's rule. If a calculation does not appear in your executive summary or in the defense of one of its two numbers, do not spend clock time on it. The pack is full of correct arithmetic that does not earn points.
Three slides, executive summary first
A tight three-slide core is enough for almost every invitation. More slides usually mean you are showing work instead of a call.
Slide 1 — Executive summary. Headline is the recommendation in one sentence. Body: the two numbers that carry it, one risk that would flip you, and the first next step. Example shape: "Close the 12 suburban branches; keep the eight urban ones. Suburban contribution is −$1.2M; urban is +$4.8M. Risk: if digital deposits stall below 8% growth, keep two suburban hubs. Next: 90-day deposit migration pilot on three sites."
Slide 2 — Evidence. The two or three charts or tables that produce those numbers. Headlines are insights ("Suburban branches lose $100k each; urban clear $600k"), not titles ("Branch P&L").
Slide 3 — Risks and next steps. One assumption that would change the call, how you would know within 90 days, and who owns the first action. Leave blank space; partners read white space as confidence.
Worked example: Harborline Bank branch closures
Invented client, invented numbers, internally consistent. Prompt: Harborline, a regional bank with 20 branches, asks whether to close underperforming branches and shift customers to digital. You have 60 minutes and three blank slides.
Exhibit log (titles only). Ex1: five-year deposit growth. Ex2: branch P&L by site. Ex3: customer digital adoption by segment. Ex4: competitor branch density. Ex5: staffing ratios. Ex6: survey on branch importance. Ex7–12: regulatory and brand fluff. Priority: Ex2, Ex3, Ex1. Medium: Ex4, Ex6. Skip: Ex5 and fluff unless Ex2 needs a cost driver.
From Ex2 (deep read). Eight urban branches: average deposits $180M, contribution +$600k each → +$4.8M. Twelve suburban: average deposits $55M, contribution −$100k each → −$1.2M. Closing all suburban removes the −$1.2M drag and frees about $2.4M of fixed cost (facilities + branch staff) if 70% of suburban deposits migrate digitally or to urban sites within a year.
From Ex3. Digital-only customers already generate 40% of new deposit dollars and cost roughly one-third of branch servicing cost per account. Suburban customers are 55% digital-ready by survey; urban customers are 70%.
Math that carries the call. (1) Closing 12 suburban branches removes −$1.2M contribution drag and, at 70% deposit retention, preserves about $462M of $660M suburban deposits ($55M × 12 × 0.7). (2) Retained deposits at urban economics of about $600k / $180M ≈ 0.33% contribution on deposits imply roughly $1.5M of contribution from migrated deposits ($462M × 0.0033), plus the $1.2M drag removed, before one-time closure costs of about $3M (severance and lease exits). Year-one P&L swing ≈ $1.2M + $1.5M − $3M amortized ≈ +$1.7M if closure costs are spread over three years ($1M/year), or near flat if you expense all $3M in year one. Recommendation still holds on a three-year view: drag gone, deposits mostly retained, digital cost-to-serve lower.
Slide text.
Slide 1 headline: "Close all 12 suburban branches; keep the eight urban ones." Body: "Removes −$1.2M annual drag; keeps $462M deposits at 70% retention. Year-one call is still positive if closure costs ($3M) are spread over three years. Flip risk: retention below 50%. Next: three-site migration pilot in 90 days."
Slide 2: table of urban vs suburban contribution; one bar of digital readiness 55% vs 70%.
Slide 3: "If digital deposit growth falls below 8%, keep two suburban hubs as deposit catchers. Owner: retail banking lead; decision gate at day 90."
How partners probe written work (and how to defend a round)
After you present, partners attack the written product, not your personality. Typical probes: "Where is the $1.2M from?" "Why 70% retention?" "What happens if retention is 40%?" "Why did you ignore Exhibit 4?" "Walk the math without looking at the slide."
Prepare for three moves. Source: name the page and the cell path ("Exhibit 2, suburban column, contribution row, twelve sites"). Sensitivity: state the threshold that flips you ("below 50% retention I keep hubs"). Rounding defense: show the path, not an apology. If you used 0.33% contribution on deposits from $600k / $180M, say "six hundred over one-eighty is one-third of a percent; I rounded to 0.33 and applied it to $462M, which is about $1.5M." Partners forgive clean rounding. They punish numbers that cannot be reconstructed.
If you skipped an exhibit, say why in one sentence: "Competitor density does not change the contribution math on our own P&L." That is judgment. "I ran out of time" is not.
Practice this today
Print any 15–25 page public deck (an annual report excerpt works). Give yourself 60 minutes. Produce three slides with an executive summary that states a recommendation and two numbers. Then have a friend ask only: "Where does number one come from?" and "What flips you?"
On CoachNed, drill the two skills the pack rewards most: chart triage on /drills/charts and closing sentences on /drills/synthesis. Everything is open for seven days, no card; then $120 for a recruiting season or $49 a month. Pair that with Bain's and BCG's published case material linked above so your voice matches what the firms train candidates to show.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named here.
Frequently asked questions
Does every BCG or Bain office use a written case?
No. Both firms run live cases as the default, per their careers pages. Written packs appear in some offices and later rounds; your invitation is the only reliable signal. Do not assume the format from online folklore alone.
How long is a written case interview?
Invitations vary. Candidates often report roughly an hour at some Bain offices and longer solo time at some BCG offices, then presentation and Q&A. Use the minutes on your invite and the 60- or 90-minute budget table above; do not memorize a single global number.
Can I use a calculator?
Follow the invitation. Some rooms allow a calculator; some do not. Practice both: clean mental math with rounding you can defend, and calculator speed when it is permitted. Never put a number on a slide you cannot rebuild aloud.
How many slides should I build?
Three is the working default: recommendation, evidence, risks and next steps. If the firm gives pre-filled templates, fill those templates rather than inventing a longer deck. Extra slides rarely add score and often dilute the call.
What is the difference between a written case and Strategy&'s structured pack?
Strategy& UK publishes a structured case where you get a slide pack and about ten minutes to review before recommending, with the interviewer present. A classic BCG or Bain written case gives you longer solo time, then a presentation. The triage and answer-first habits transfer; the clock does not.
Sources
- BCG Case Interview Preparation — what BCG publishes about case interviews (structure, analysis, communication). Checked 2026-09-24.
- Bain Interviewing — Bain's public description of consulting interviews, assessments, and written assignments. Checked 2026-09-24.
- Bain FashionCo practice case — Bain's published good vs better recommendation standard with a dollar gap. Checked 2026-09-24.
- Strategy& UK Careers FAQs — structured case with a client-style slide pack and ~10 minutes to review. Checked 2026-09-24.