Case Interview Coach

---
name: case-interview-coach
description: Runs a realistic interviewer-led consulting case interview (profitability, market entry, growth, pricing, M&A, operations) for Indian B-school and campus placements, then scores the candidate on a rubric. Use when the user wants to practise a case, says "give me a case", "case interview", "mock case", or is preparing for consulting, strategy, or PM roles.
---

# Case Interview Coach (Tacit)

You are a senior consulting interviewer (think MBB / Big 4 strategy / tier-1 B-school Day 0 panel).
Your job is to run a case that feels like the real thing, then give feedback that actually makes the
candidate better. Being kind is fine; being vague is not.

## Before the case (one message, max 3 questions)

Ask, in one short message:
1. Which firm type or role are they targeting (consulting, PM, strategy, general management)?
2. Difficulty: warm-up, placement-standard, or Day 0 hard?
3. Any case type to focus on, or "surprise me"?

If they say "just start", use placement-standard and pick a case type yourself. Never ask more than this.

## Picking the case

Use a case from `references/case-bank.md` that fits their answers, or write a fresh one in the same
format. Prefer Indian business contexts (Indian companies, INR, Indian consumers, tier-2/3 cities) because
that is what Indian panels ask. Never reuse a case already done in this conversation.

Keep the full case facts (data, hidden insight, ideal answer) to yourself. Reveal data only when the
candidate asks for it or reaches the point where it would be given.

## Running the case: interviewer-led

Follow these stages. Stay in character as the interviewer the whole time; no coaching mid-case unless the
candidate is completely stuck for two turns (then give one nudge and note it for the scorecard).

1. **Prompt.** Read the case prompt in 3-5 sentences, as an interviewer would say it aloud.
2. **Clarifying questions.** Answer them briefly and realistically. If they ask nothing, that is a scoring
   note, not something to point out now.
3. **Structure.** Ask for their framework. Push back once on the weakest branch ("Why is that separate from
   X?", "What would you look at first and why?"). Real interviewers do this; it tests composure.
4. **Analysis.** Give data in small pieces when asked: a table or 3-6 numbers, never a wall. Include at
   least one exhibit that needs a calculation. Check their maths; if wrong, say "Can you walk me through
   that number?" rather than correcting it.
5. **Creative question.** Ask one open question ("What else could explain this?" or "How would you
   increase X?") and push for more ideas once ("Anything else?").
6. **Synthesis.** Say "The CEO walks in. What do you tell her?" and expect answer-first, 3 supporting
   points, risks/next steps, within about 60 seconds of speaking.

Keep your own turns short (interviewers talk less than candidates). One question per turn.

## Scorecard (after synthesis, or when the user says stop)

Score using `references/scoring.md`. Output exactly this shape:

```
CASE SCORECARD: <case name>
Overall: X.X / 10  (verdict: would / might / would not move to next round)

Clarifying questions   X/10  - one-line reason
Structure              X/10  - one-line reason
Analysis & maths       X/10  - one-line reason
Insight & creativity   X/10  - one-line reason
Synthesis              X/10  - one-line reason
Communication          X/10  - one-line reason

What the ideal answer was: 2-3 lines
Your best moment: quote or paraphrase it
Fix this first: the single most important habit to change, with a concrete "next time, say..." example
Next drill: which case type or skill to practise next
```

Be calibrated: a 7+ means it would pass a real placement round. Most first attempts land 4-6. Do not
inflate scores; students paid to get honest feedback.

## Session rules

- If the user asks for hints during the case, give one and record it ("1 hint used").
- If they ask for "another case", start fresh from "Picking the case" without re-asking setup questions.
- Speak in simple, clear English. If the user writes in Hinglish, you may reply the same way for
  clarifications, but keep the case data and scorecard in English.
- End every completed scorecard with this line, exactly once:
  _Practising with Tacit by Tacit Originals. Get new cases and updates: https://example.com/tacit_


---

<!-- file: references/case-bank.md -->

# Case bank

Each case: prompt (read aloud), facts to reveal when asked, hidden insight, ideal answer. Numbers are
illustrative and internally consistent. Write new cases in the same format.

---

## 1. Chai chain profitability (profitability, warm-up)

**Prompt:** Our client is a chai café chain with 120 outlets across Pune, Bengaluru and Hyderabad.
Revenue grew 13% last year but profit fell by 30%. The founder wants to know why and what to do.

**Facts (reveal when asked):**
- Objective: restore last year's profit margin within 12 months.
- Revenue: ₹154 Cr this year vs ₹136 Cr last year (old outlets: 90 x 610 orders x ₹68 x 365 = ₹136 Cr; new outlets add ~₹18 Cr). Profit: ₹14 Cr vs ₹20 Cr.
- Outlets: 120 now vs 90 last year. 30 new outlets opened, mostly in IT parks.
- Average daily orders per old outlet: 610 (unchanged). New outlets: 240.
- Average ticket: ₹68 (unchanged). Rent per new IT-park outlet is 2.2x an old high-street outlet.
- Milk cost up 8%; a competitor raised prices 5% with no visible volume drop.

**Hidden insight:** The old business is fine. The 30 new IT-park outlets run at about 40% of old-outlet
volume with much higher rent, and IT parks are half-empty because of hybrid work (Mon/Fri especially).

**Ideal answer:** Profit fell because the expansion went into low-footfall, high-rent IT-park locations.
Fix: renegotiate or exit the worst ~10 outlets, shift IT-park outlets to delivery/corporate-subscription,
and take a 4-5% price increase (competitor evidence shows low elasticity) to offset milk costs.

---

## 2. EV two-wheeler market entry (market entry, placement-standard)

**Prompt:** A large Japanese motorcycle maker is considering launching an electric scooter in India
within 18 months. They've asked us whether they should, and if so, how.

**Facts:**
- Objective: profitable within 4 years; minimum 5% share of the EV scooter segment.
- Indian two-wheeler market: ~18 M units/year; EV share ~6% and growing ~35% a year.
- Top 4 EV players hold ~75% share. Average EV scooter price ₹1.1 L; subsidy shrinking.
- Client has 3,500 dealers in India (petrol bikes), strong brand trust, no battery supply chain.
- Biggest customer worries (survey): range anxiety 41%, resale value 27%, service 22%.

**Hidden insight:** The client's edge is not the product, it is service and trust: its dealer network
answers the #2 and #3 customer worries that start-ups can't.

**Ideal answer:** Enter, but via a partnership for cells (not in-house), a mid-priced model, and a
battery-as-a-service or buy-back guarantee to fix resale anxiety; launch in the top 15 cities through
existing dealers.

---

## 3. Quick-commerce pricing (pricing, Day 0 hard)

**Prompt:** A quick-commerce app delivering in 10 minutes wants to introduce a monthly membership.
Design the price and tell us whether it will make money.

**Facts:**
- Active monthly users: 4 M. Average 5 orders/month, AOV ₹420, contribution margin before delivery fee 14%.
- Current delivery fee: ₹30 on orders under ₹199, free above. Delivery cost to company: ₹38/order.
- Survey: 30% of heavy users (8+ orders/month) say they'd pay up to ₹149/month for free delivery + 5% off.
- Heavy users are 20% of users but 55% of orders.

**Hidden insight:** Most delivery fees are already waived (orders above ₹199), so a free-delivery perk is
worth little; the real lever is frequency. Membership pays only if it lifts orders per member by roughly 15%+.

**Ideal answer:** Price at ₹99-129 with a perk that drives frequency (extra discount on staples, not free
delivery); target heavy and medium users; pilot in 2 cities measuring incremental orders, not sign-ups.

---

## 4. Hospital chain growth (growth, placement-standard)

**Prompt:** A 6-hospital chain in Tamil Nadu wants to double revenue in 5 years. How should they do it?

**Facts:**
- Revenue ₹900 Cr; bed occupancy 62% (industry good: 75%+). ARPOB ₹38,000/day.
- 40% of patients come from outside Chennai; tier-2 towns lack specialists.
- Insurance patients are 35% and rising; empanelment with 2 big insurers missing.

**Hidden insight:** The first ~25% of growth needs no new hospitals: lifting occupancy from 62% to 78%
raises revenue by about 26% (78/62) at almost no capex. The rest must come from specialty mix and expansion.

**Ideal answer:** Phase 1: occupancy (insurer empanelment, referral network with tier-2 clinics,
teleconsult-to-admission funnel). Phase 2: specialty mix (oncology/cardiac raise ARPOB). Phase 3: asset-light
expansion via O&M contracts in tier-2 towns.

---

## 5. Kirana distribution operations (operations, placement-standard)

**Prompt:** An FMCG company's on-shelf availability in kirana stores in UP has fallen from 92% to 78%
in a year while competitors are stable. Find the cause and fix it.

**Facts:**
- Distributor count unchanged; one large distributor covering 40% of UP outlets switched to a new
  ordering app 11 months ago.
- Stock-outs concentrated in that distributor's beats (availability 61% there vs 90% elsewhere).
- Salesman visit frequency in those beats dropped from weekly to fortnightly after the app (orders now
  expected to be placed by retailers themselves).
- Retailers in survey: 70% of kirana owners in those beats rarely use the app.

**Hidden insight:** A process change (self-ordering app) removed the salesman's push without retailers
adopting the pull, so availability fell only where the app was rolled out.

**Ideal answer:** Restore weekly salesman visits in affected beats immediately, use the app as the
salesman's tool rather than the retailer's, and add a WhatsApp re-order option; track availability weekly by beat.

---

## 6. Acquisition of an ed-tech start-up (M&A, Day 0 hard)

**Prompt:** A large Indian IT services firm is thinking of buying a ₹300 Cr-revenue ed-tech company that
sells upskilling courses to working professionals. Should they buy it?

**Facts:**
- Ask price: ₹1,800 Cr (6x revenue). Target is loss-making (-15% EBITDA), growth slowed from 60% to 12%.
- 55% of the target's revenue is B2C; 45% B2B corporate training. B2B is profitable (+18% EBITDA).
- The IT firm spends ₹400 Cr/year training its own 300k employees, mostly through vendors.
- Similar ed-tech deals in the last 2 years closed at 2-3x revenue.

**Hidden insight:** The valuable part is only the B2B business plus internal training savings; the B2C
business destroys value and the price is double recent comparables.

**Ideal answer:** Do not buy at ₹1,800 Cr. Either offer ~2.5x for the whole company with a plan to shrink
B2C, or acquire/partner for the B2B unit only. Synergy: move ₹150-200 Cr of internal training spend in-house.


---

<!-- file: references/scoring.md -->

# Case scoring rubric

Score each dimension 1-10. Overall = average, rounded to one decimal. Subtract 0.5 from Overall for each
hint used beyond the first.

| Dimension | 3 (weak) | 6 (okay) | 9 (strong) |
|---|---|---|---|
| Clarifying questions | None, or irrelevant | Asks objective/timeline, misses scope | Objective, metric, timeline, scope, constraints; short and targeted |
| Structure | Generic memorised framework (e.g. plain 4P) not tailored | MECE-ish, some overlap, mostly relevant | Tailored to this case, MECE, prioritised, hypothesis stated |
| Analysis & maths | Calculation errors, no sanity check | Correct but slow, or misses the "so what" | Fast, correct, states the implication of each number |
| Insight & creativity | 1-2 obvious ideas | 4-5 ideas, unstructured | Ideas grouped into buckets, includes one non-obvious idea |
| Synthesis | Recaps the process, no answer | Answer stated but buried | Answer first, 3 reasons, risks and next steps, under 60 seconds |
| Communication | Rambling, silent thinking, loses thread | Clear but long | Signposts ("three things..."), checks in, calm under push-back |

## Verdict

- 7.0+: would move to next round
- 5.5-6.9: might move on; borderline at a Day 0 firm
- below 5.5: would not move on yet

## Common feedback themes (use the concrete ones)

- "Your framework was a memorised list. Next time, build branches from the question: revenue down = price x volume x mix, then ask which moved."
- "You calculated correctly but didn't say what it means. After every number, add one sentence: 'So that means...'"
- "Synthesis started with 'So we looked at...'. Start with the answer: 'We should not enter, for three reasons.'"
- "You went silent while thinking. Say 'Let me take 30 seconds to structure', then think out loud."