A written case swaps your interviewer for a stack of slides and a clock. Instead of talking through a problem in real time, you get 20-40 pages of data, roughly an hour to build a recommendation, and 15-20 minutes to present it to a panel that has never heard your reasoning out loud. Firms use this format specifically because it isolates a skill spoken cases can’t test: can you find the signal in a pile of noisy information fast enough, then write something a partner could hand to a client without editing it?
Why Firms Test Written Delivery Separately
Bain runs written cases heavily in its European and Greater China offices, reserving them for final rounds alongside a fit interview. You’ll typically get 20-30 slides describing a client situation, 55 minutes to extract the key data and complete 3-5 partially-filled slides by hand, then 40 minutes to present and defend your analysis — without a calculator. LEK and McKinsey run similar formats in select markets, though slide counts and time limits vary by office.
The scoring rubric mirrors a normal case (structuring, analysis, business judgment) but weights two things a live case can’t isolate:
- Signal extraction under volume: you’re deliberately given more data than you need, so triage speed is being graded, not just math accuracy.
- Written persuasion: your slides have to carry the argument without you narrating every step, because the panel reads them cold before you speak.
The 5R Structure That Replaces the Standard Case Flow
Standard spoken cases move opening → structure → analysis → conclusion, building up to the recommendation at the end. Written cases invert this. Lead with your answer first, then support it — because a reader scanning slides guesses where you’re headed anyway, and you want that guess to be right from the first page.
flowchart LR
A[Recap the Question] --> B[State the Recommendation]
B --> C[Show the Reasoning]
C --> D[Flag the Risks]
D --> E[Retain: Propose Next Steps]
- Recap the question. One slide, one sentence. It forces you and the panel onto the same target before anyone reads a number.
- State the recommendation. Lead with the answer, not the buildup. A reader who has to guess your conclusion stops absorbing your evidence and starts predicting it — often wrong.
- Show the reasoning. One slide per supporting pillar, each with the data that backs it. This is where your case math has to be airtight, since there’s no interviewer to catch an error mid-calculation.
- Flag the risks. Most candidates stop at a correct recommendation. Naming what could break it — market shift, execution risk, competitor response — is what separates a pass from a strong pass.
- Retain the client. Close with concrete next steps that show how your team would keep working the problem. It signals you understand that consulting is a relationship, not a single deliverable.
Message Headings, Not Chart Titles
The single biggest quality gap between a junior and senior-looking deck is the slide title. A chart title describes the data. A message heading states the implication a reader should walk away with.
| Chart title (weak) | Message heading (strong) |
|---|---|
| “Product C sales by region” | “Product C’s strength in the domestic market makes it the best candidate for regional expansion” |
| “Cost breakdown by category” | “Labor costs, not materials, are driving the 12-point margin decline” |
| “Customer churn over time” | “Churn accelerated after the Q3 price increase, pointing to a pricing — not product — problem” |
Every slide in your reasoning section should pass a simple test: could someone understand your argument by reading only the headings, without looking at a single chart? If yes, your deck is doing its job.
Time Allocation for a 55-Minute Prep Window
| Phase | Time | What to do |
|---|---|---|
| Skim and triage | 10-12 min | Read every slide title first, then flag which 30% of the data actually answers the question |
| Deep dive on key data | 20-25 min | Do the calculations that support your recommendation only — skip analysis that doesn’t move the answer |
| Draft slides | 15-18 min | Fill in the 3-5 provided slides (or build your own) using the 5R order, message headings throughout |
| Rehearse the walk-through | 3-5 min | Decide what you’ll say for each slide in under 60 seconds — panels penalize decks that need a live narrator to make sense |
Running out of time in the skim phase is the most common failure mode. If you’re still reading slide 25 with 10 minutes left, you’ve already lost — cut your losses, commit to the strongest data you’ve seen, and move to drafting.
Delivery: Defending a Deck You Can’t Improvise Around
The presentation half of a written case has a different pressure profile than a live case. You already committed your logic to paper, so the panel’s questions will often probe exactly the gaps you knew about but didn’t have time to close. Two habits matter here:
- Open with your recommendation verbatim from slide one, then walk the reasoning in order. Don’t re-derive your answer live — that suggests you didn’t actually reach it during prep.
- When challenged on an assumption, don’t retreat to “I didn’t have time.” Say what you’d check next and why, which reframes a gap as a next step rather than an oversight.
Practicing the Skill Without a Real Deck
Most candidates prepare for spoken cases and only discover the written format’s different rhythm during the actual interview — by then it’s too late to build the “lead with the answer” instinct. CasesCoach’s Chat with Case mode gives you a lower-stakes way to build the same muscle: you work through a real business problem in five stages — background, framework selection, data, analysis, and conclusion — which mirrors exactly the sequence a written deck needs to compress into slides. Practicing that flow repeatedly makes the 5R structure feel automatic instead of something you’re reconstructing under a countdown timer.
For the live defense half, Expert Mode in the AI Mock Interview simulates the pressure of a panel probing your logic in real time, while Guided Mode is the better starting point if you’re still learning to structure a recommendation before adding time pressure. Every mock session scores your performance across five dimensions — Structure, Analysis, Communication, Business judgment, and Math — so you can see precisely whether your reasoning or your delivery is the weaker link.
Key Takeaways
- Written cases test signal extraction and written persuasion — skills a spoken case can’t isolate, which is exactly why firms like Bain and LEK use both formats.
- Invert the standard case flow: lead with your recommendation first, then support it with the 5R structure (Recap, Recommendation, Reasoning, Risks, Retain).
- Replace chart titles with message headings — every slide should carry its conclusion in the heading, not just in the data.
- Budget your prep time explicitly: skim and triage should take no more than 20% of your available time, or you’ll run out before you draft.
- Practice the underlying five-stage reasoning flow with Chat with Case, then stress-test your live defense with AI Mock Interview before the real thing.
Ready to build the instinct instead of just reading about it? Start with three free full cases and an AI Mock Interview — no card required. When you’re ready for the full 835+ case library and multiple Mock attempts, Pro unlocks everything.