To size a drug market in a case interview, build a patient funnel: population, times prevalence or incidence, times diagnosis rate, times treatment rate, times your share, times adherence, times annual net price. State the whole structure out loud before you touch a number. The scoring is in the logic and the assumptions you choose to defend, not in the arithmetic.
The funnel, step by step
Every life-sciences sizing question runs through some version of this shape. Learn the shape, not a specific example.
| Step | What it is | What it does to the answer |
|---|---|---|
| Population | The relevant geography, usually US adults | Sets the ceiling. Say which geography you are in. |
| x Prevalence or incidence | People living with the disease, or new cases per year | The single biggest swing factor in the funnel. |
| x % diagnosed | Patients the health system has actually identified | Often 30 to 50% of patients are undiagnosed. Skipping this overstates badly. |
| x % treated | Diagnosed patients who get drug therapy at all | Filters out watch-and-wait and non-pharmacologic management. |
| x your share | Peak share of the treated pool | In a crowded class, 15 to 25% is often realistic. |
| x adherence | Patients who actually stay on therapy | Real-world persistence on chronic therapy is often 50 to 80%, not 100%. |
| x annual net price | Dose schedule times price, less gross-to-net | Turns patients into revenue. The step most candidates fumble. |
Say the structure before you calculate. "I'll size the eligible population through a patient funnel, apply a net price, then pressure-test diagnosis rate and share, which I think swing the answer most." Now the interviewer can redirect you early instead of watching you scribble.
Prevalence or incidence, and why this one choice decides the case
Prevalence is everyone living with the condition at a point in time. Incidence is new cases per year. Use prevalence for chronic diseases where patients stay on therapy. Use incidence for acute or one-time treatment, such as a cancer treated in a defined course or a single-administration gene therapy.
Getting this backwards is the most common sizing error, and it is not a rounding error. A chronic drug sized off incidence undercounts the market by every year of accumulated patients still alive and still treated. An acute therapy sized off prevalence does the reverse.
The trap nobody warns you about: which prevalence number
Even with the right concept, the number itself can be ambiguous by nearly an order of magnitude, because survey-reported disease and clinically confirmed disease are not the same population.
Rheumatoid arthritis is the cleanest illustration. Analysis of NHANES 2017 to March 2020 data put diagnosed arthritis at 27.9%, or 67.1 million US adults aged 20 and over, and reported that 15.8% of those adults had rheumatoid arthritis, which works out to roughly 10.6 million adults, an age-standardized 4.1%. That is a self-reported, survey-based figure.
The long-standing clinical and registry literature puts rheumatoid arthritis far lower, commonly cited in the range of roughly 0.5% to 1% of adults. Both numbers are real. They measure different things, because self-report in a household survey picks up osteoarthritis and other joint disease that respondents call rheumatoid.
Pick the conservative number, label it, and say why. "I'll use roughly 1% of adults, which is the clinically confirmed range. Survey data reports several times higher because it is self-reported, and for a therapy that needs a confirmed diagnosis to get reimbursed, the confirmed population is the one that can actually be sold to." That sentence does more for you than three decimal places.
Worked example, illustrative throughout
A chronic autoimmune condition, oral therapy, crowded class. Every percentage below is illustrative and chosen to be roughly realistic, not researched for this specific disease.
| Step | Assumption | Running number |
|---|---|---|
| US adults | Roughly 260 million | 260,000,000 |
| Prevalence | 1% [illustrative, conservative end of the confirmed range] | ~2.6 million |
| Diagnosed | 60% [illustrative] | ~1.6 million |
| On drug therapy | 70% [illustrative] | ~1.1 million |
| Your peak share | 20% [illustrative, crowded class] | ~220,000 |
| Adherent at steady state | 70% [illustrative] | ~155,000 |
| List price per year | $30,000 [illustrative] | |
| Less gross-to-net | 25% [illustrative, crowded class] | $22,500 net |
| Peak net revenue | ~$3.5 billion |
Now sanity-check, out loud. Roughly $3.5 billion is a large but plausible peak for a specialty product in a chronic autoimmune indication. If the funnel had produced $50 billion, something upstream is wrong. If it had produced $50 million, the honest answer is that this asset may not clear a launch hurdle, and saying so is a stronger finish than a confident wrong number.
Net price is where candidates lose points quietly
Two separate adjustments sit between a price and a revenue number, and skipping either one produces an answer that is wrong by a multiple rather than a rounding error.
Dosing schedule. A vial that costs $5,000 and is administered four times a year is a $20,000 annual cost, not $5,000. Infused and injected products are priced per administration, so the annualization step is doing real work.
Gross-to-net. List price is not what the manufacturer collects. Rebates to payers and pharmacy benefit managers, copay assistance, and channel discounts all sit in between. A feasible range for a case is roughly 5 to 50%, driven by how competitive the class is and how hard payers can gate it. When you have nothing to anchor on, 20% is a defensible default. Name it as an assumption and keep moving.
The funnel changes shape by modality
Reaching for the drug funnel on a device case is one of the most common ways to lose an interview, and it is entirely avoidable.
| Modality | What you size instead |
|---|---|
| Drug | The funnel above, times annual net price. |
| Device | Eligible procedures per year, times device use rate. There is no adherence on an implant. For capital equipment, size the install base and the replacement cycle. |
| Diagnostic | Testable population, times testing rate, times your share of tests. Screening and monitoring tests repeat, often annually, so they behave like a recurring funnel. Watch the ordering-rate gate. |
| Digital health | Covered or enrolled lives, not disease prevalence. Revenue is per-member-per-month times enrolled members, gated by contracts and engagement. |
Six traps, in the order they usually show up
- Prevalence and incidence mixed up. Covered above. It invalidates everything downstream.
- Diagnosis rate skipped entirely. Going straight from disease population to treated patients quietly assumes the health system finds everyone, which it does not.
- Adherence set to 100%. Treating every prescription as a patient on therapy for a full year overstates a chronic market by a third or more.
- List price used as net price. Plus forgetting to annualize the dose.
- False precision. "6,048,000 patients" tells the interviewer you do not understand that these are estimates. Round, and round visibly.
- No sanity check. Ending on a number with no reaction to it is the single most common way a technically correct case still reads as junior.
How the four life-sciences boutiques use this
Sizing is table stakes at all of them, which means it is necessary and never sufficient. What sits on top differs by firm, and that is where the offer is decided.
At ClearView the funnel usually feeds a pipeline or portfolio prioritization question, alongside a separate science discussion. At Putnam it feeds pricing and market access, where coverage, coding and reimbursement decide whether your sized market is reachable at all. At ZS the analogous drill is sales force sizing, built on prescriber concentration rather than patient counts. At Trinity the funnel shows up inside launch and forecasting cases, with fit scored in the same interview. The firm-by-firm comparison lays out the loops side by side.
If you came to sizing from a bench or clinical background, the companion problem is talking about the science without over-indexing on it. That is a separate skill and a separate round at ClearView.
Frequently asked questions
How do you size a pharmaceutical market in a case interview?
Build a patient funnel. Population, times prevalence or incidence, times diagnosis rate, times treatment rate, times your share, times adherence, times annual net price after dosing and gross-to-net. State the structure before you calculate anything.
Should I use prevalence or incidence?
Prevalence for chronic conditions where patients stay on therapy. Incidence for acute or one-time treatment. Sizing a chronic drug off incidence undercounts by years of accumulated patients, and it is the most common error candidates make.
What is a gross-to-net discount and do I need it in a case?
It is the gap between list price and what the manufacturer actually collects after rebates, copay assistance and channel discounts. You need it, because revenue is built on net price. Roughly 5 to 50% is the feasible range; 20% is a reasonable default to state and move past.
Does the patient funnel work for devices and diagnostics?
Not as written. Devices are sized on eligible procedures, diagnostics on testable population times testing rate, digital health on enrolled lives. Practice at least one device case before an interview loop.
How do I know if my answer is right?
You do not, and it is not what is being scored. Sanity-check against a known analog. A specialty drug peaking around $2 to $4 billion is plausible; $50 billion means an upstream error; $50 million is itself a finding.
Drill it cold
Work 5 life-sciences cases free
Five prompts across drugs, devices, diagnostics and digital health, plus the patient funnel framework on one page. Work them before you look at the hints.