I stop scoring when a candidate multiplies prevalence by wholesale acquisition cost and calls it a forecast. That product is a press release, not a P&L. In manufacturer-side pharma cases, the interviewer is listening for three clocks: how patients leak through a clinical funnel, how much cash the firm actually keeps after rebates, and how many years of exclusivity remain.
The thesis is blunt. A pharma case is a discounted cash flow until loss of exclusivity, not a hospital bed case and not a snack-brand share defense. Volume is diagnosed × treated × share × days on therapy. Price is net after the gross-to-net stack. Peak sales after the patent dies are decoration. After this piece you should be able to build a year-3 launch forecast from prevalence to net revenue, stress a gross-to-net haircut, and refuse a list-price recommendation that ignores the cliff.
This is the manufacturer lane. Provider operations belong on the healthcare case interview page. Devices, diagnostics, and mixed pipelines belong on life sciences. A sibling guide covers IQVIA-style commercial analytics; here the interview math stays firm-agnostic.
What the room actually rewards
In a typical first round of eight, two or three candidates will open with "Porter five forces on Big Pharma." That burns two minutes and earns nothing. The score sheet rewards a decision with arithmetic:
| Decision the prompt is hiding | Number that settles it |
|---|---|
| Launch / delay / kill | Year-1 to year-3 net sales vs remaining exclusivity |
| Price vs access | Net price × realized share after step-edit, not WAC × "fair share" |
| Indication sequence | Incremental patients and incremental patent life, not vanity labels |
| Pre-LOE harvest | Cash until cliff vs post-LOE salvage (authorized generic, next-gen) |
Clarify brand vs generic vs biosimilar, US vs EU5 vs rest of world (US net usually dominates value), specialty vs primary care (sales-force math changes), and whether every dollar on the exhibit is WAC, invoice, or net. If you cannot name which, ask.
The industry numbers candidates should actually cite
FDA's Center for Drug Evaluation and Research approved 50 novel drugs in 2024 — new molecular entities and therapeutic biologics never before marketed in the US. The agency's own 2024 approvals report puts the ten-year average near 47 per year. Approval is not access: a drug can clear FDA and still sit behind a payer step-edit for a year.
Clinical-phase success is worse than folklore admits, and the provenance matters. BIO / Informa Biomedtracker's Clinical Development Success Rates 2011–2020 tracked 12,728 phase transitions across 9,704 programs: Phase I to II 52.0%, Phase II to III 28.9% (the choke point), Phase III to NDA/BLA 57.8%, compounding to a 7.9% likelihood of approval from Phase I. An independent academic estimate (Wong, Siah, and Lo, Biostatistics, using Informa data 2000–2015) put overall Phase I-to-approval nearer 13.8% — same industry, different sample and method. In the room, cite the source and the year; do not invent "one in ten" as if it were a law of nature.
Patent cliffs show up in 10-Ks, not in memes. AbbVie's 2024 Form 10-K discloses global Humira net revenues of $21.237 billion (2022), $14.404 billion (2023), and $8.993 billion (2024) after US biosimilar entry — roughly a 58% drop from the pre-LOE peak in two years, while successor brands carried the immunology book. That is what "volume holds, price does not" looks like in filing language.
List price is a fiction for profitability. An NBER working paper on US retail pharmacy brands estimated average rebates rose from about 32% of list in 2012 to 48% in 2017. Manufacturers report net sales; the rebate stack (PBM, Medicaid best price, 340B, copay offset) sits between WAC and cash. Under the Inflation Reduction Act, CMS also negotiates a maximum fair price for selected Medicare drugs and requires manufacturers to submit US market and revenue data — another reason "just raise US list price" is not a strategy.
The funnel that replaces TAM theatre
Prevalence is not treated patients. Write the leak, then defend each rate:
Eligible treated = Prevalence × diagnosis rate × treatment rate × eligible share
Brand patients = Eligible treated × brand share × persistency (or months on drug / 12)
Net revenue = Brand patients × net price per patient-year
Oncology often sizes patients × duration × vials. Primary-care diabetes sizes scripts × days' supply. Do not mix units. Analog launch curves usually peak in year 4–6, not year 1 — front-loading year-1 share to "hit the CFO number" is a classic miss.
Gross-to-net is a bridge, not a rounding error. A 9% list increase can be a 2% net decline if rebates and copay cards catch up. If the exhibit gives WAC, ask for the net haircut before you forecast.
Ned's rule. Never quote a pharma revenue number that still contains WAC or untreated prevalence. Convert both before you recommend. If the recommendation only works on list price or on "all diagnosed patients," it fails the case.
Worked mini-case: Auralyn year-3 launch forecast
Prompt (invented; numbers consistent, not real). Northwind Pharma has FDA approval for Auralyn, an oral specialty drug for moderate-to-severe atopic dermatitis in US adults. Management wants a year-3 US net sales forecast and a go/no-go on a $185 million commercial build (specialty sales force, payer contracting, samples, medical affairs). Exclusivity runs 8 years from today. Competing injectables already own most of the treated pool. Case inputs:
| Input | Value |
|---|---|
| US adult prevalence (moderate-severe AD) | 8.0 million |
| Diagnosed | 55% |
| Of diagnosed, currently treated with Rx systemic or specialty | 40% |
| Of treated, clinically eligible for Auralyn label | 70% |
| Year-3 brand share of eligible treated (base) | 12% |
| Average months on therapy in year 3 | 9 of 12 (persistency 0.75) |
| WAC per patient-year | $28,000 |
| Gross-to-net (rebates, fees, copay offset, 340B mix) | 42% of WAC |
| Variable COGS + distribution per patient-year | $1,800 |
| Medical + commercial variable cost per patient-year | $2,200 |
Step 1 — Treated funnel, not prevalence
Diagnosed = 8.0m × 0.55 = 4.40m
Currently treated = 4.40m × 0.40 = 1.76m
Eligible treated = 1.76m × 0.70 = 1.232m
Year-3 brand pts = 1.232m × 0.12 × 0.75 ≈ 110,900 patients
If someone stops at "8 million times 12% share," they just invented ~960,000 patients. That is nearly nine times the funnel. I mark that as a structure fail even if the arithmetic is tidy.
Step 2 — Net price, then contribution
Net price / patient-year = $28,000 × (1 − 0.42) = $16,240
Contribution / patient = $16,240 − $1,800 − $2,200 = $12,240
Year-3 net sales ≈ 110,900 × $16,240 ≈ $1.80 billion
Year-3 contribution ≈ 110,900 × $12,240 ≈ $1.36 billion
The $185 million build is about 14% of year-3 contribution in the base case — not the constraint. The constraint is whether year-3 share of 12% is earned with a step-edit behind two injectables. Cut share to 7% and patients fall to ~64,700; net sales fall to ~$1.05 billion. Raise the gross-to-net haircut to 50% (net $14,000) at 12% share and sales fall to ~$1.55 billion. Both sensitivities belong on the board.
Step 3 — Clock it against exclusivity
Eight years of exclusivity means the beautiful year-10 peak is mostly generic or biosimilar economics. Rough interview sketch: assume peak year-5 net sales of $2.4 billion, then LOE year-9 with 65% of patients remaining at 22% of brand net price:
Post-LOE year net ≈ (peak patients path) × 0.65 × 0.22 × prior net
Even without perfect patient counts, a 78% price collapse with 35% volume loss is an ~86% revenue hole — the same shape as Humira's filed decline, different molecule. Do not cut net price 15% today to "defend share into the cliff"; you train every payer that the brand discounts and you worsen Medicaid best-price references. Harvest access, freeze sales-force expansion, and stand up salvage (authorized generic or next-gen with new IP) only if the cash collects before or despite LOE.
Recommendation. Go on the commercial build if payer research supports at least mid-single-digit year-3 share without buying volume with copay cards that vaporize the net. Price the brand on net willingness of specialty pharmacy and national accounts, not on WAC parity with injectables. Kill any indication expansion that spends three years of trial cash after the patent dies unless it creates a new exclusivity clock.
Where generic profitability trees break
- Time is the product. A margin win in year 9 is worthless if year 9 is generic.
- Price is net. List × volume is how candidates fail exhibits.
- Volume is clinical. Diagnosis, eligibility, prior auth, and persistency beat awareness ads.
- COGS is the wrong cost conversation. Medical affairs, pharmacovigilance, and payer contracting run the operating system; plant-style "cut cost of goods" rarely moves the case.
- Regulation sets the feasible set. FDA approval, EMA timing, IRA negotiation eligibility, and Medicaid best price constrain the moves you can recommend.
Treat Auralyn like a snack defending shelf space and you will recommend a coupon into a patent expiry and call it share.
Practice the funnel today
Run one manufacturer launch or LOE case where you refuse to use WAC or raw prevalence in the final recommendation. Force every revenue line through diagnosed → treated → share → net.
Start a five-minute scored rep (three typed turns, instant scores, no account). For a full voice case with a seven-score debrief, use /interview. Everything on CoachNed is open for seven days with no card; then $120 for a recruiting season or $49 a month. The quick math drill stays free forever if you only need unit practice.
Frequently asked questions
What is a pharma case interview testing?
Whether you can turn a molecule into net cash under a patent clock: patient funnel, gross-to-net, access, and remaining exclusivity. It is not testing hospital DRG math or CPG brand frameworks, even when the industry word "healthcare" appears in the prompt.
How is a pharma case different from a hospital healthcare case?
Hospitals live on beds, discharges, payer mix, and labor. Manufacturers live on treated patients, net price, sales force, and LOE. If the client owns a hospital system, you are on the wrong tree; start with healthcare cases instead.
What is gross-to-net in a case interview?
The bridge from list (often WAC) to manufacturer cash after rebates, fees, chargebacks, 340B, and patient assistance. Interviewers want you to forecast on net. Quoting list as if it were revenue is a common first-round fail.
Do I need drug-specific FDA knowledge for the interview?
You need the labels of the industry: approval is not reimbursement, Phase II is the usual kill point in published success-rate studies, and exclusivity bounds the DCF. You do not need to recite PDUFA calendars. If an exhibit cites an indication or a remaining patent year, use those numbers.
How should I size a drug launch in a case?
Start from prevalence only to get to diagnosed and treated, then apply eligible share, brand share, and persistency. Convert WAC to net. State peak year and LOE year. If the prompt gives analogs, borrow their uptake shape rather than inventing year-1 dominance.
Sources
- FDA Novel Drug Approvals for 2024 — CDER approved 50 novel drugs in 2024. Checked 2026-09-24.
- FDA Advancing Health Through Innovation: New Drug Therapy Approvals 2024 (PDF) — first-cycle and ten-year average context for novel approvals. Checked 2026-09-24.
- BIO / Informa, Clinical Development Success Rates and Contributing Factors 2011–2020 (PDF) — phase transition rates and 7.9% Phase I LOA. Checked 2026-09-24.
- AbbVie Form 10-K for year ended December 31, 2024 (SEC) — Humira net revenue decline after US LOE. Checked 2026-09-24.
- NBER Working Paper 26846, Rebates in the Pharmaceutical Industry — estimated rise in average list-to-net rebate share for retail brands, 2012–2017. Checked 2026-09-24.
- CMS Medicare Drug Price Negotiation Program final guidance (PDF) — IRA negotiation and manufacturer data submission context. Checked 2026-09-24.
CoachNed is independent and unaffiliated with McKinsey, BCG, Bain, or any firm named for interview-style practice.