How to credibly size a drug's addressable market (TAM) for investors

A credible drug market size is built bottom-up from epidemiology, not quoted from a market report. Start with US incidence or prevalence from a public registry, then cut it down by diagnosis rate, stage, biomarker status, line of therapy, and realistic share. Multiply by net price and real treatment duration, then check the result against what comparable drugs actually earn.

What does a credible bottom-up market model actually contain?

In my experience, the models investors trust share one structure. They begin with a population count from a named public source and apply explicit percentages, each with a citation or a labeled assumption. Every step is visible, so a reviewer can disagree with one factor without rejecting the whole model.

The funnel has a standard order. Count the people who get the disease, then those who are diagnosed, then those who reach the stage and treatment line your drug targets. Apply biomarker or eligibility filters, then an achievable share of treated patients. Finally, multiply by net price per patient and realistic duration of therapy.

Should I start from incidence or prevalence?

Incidence is the number of new diagnoses per year. Prevalence is the number of people alive on a given date who have ever been diagnosed, so it depends on both past incidence and survival [5]. Use incidence for a therapy given once or as a short course, and prevalence for a chronic therapy taken for years.

The gap can be large. For US lung and bronchus cancer, SEER estimates 229,410 new cases in 2026, while about 661,853 people were living with the disease in 2023 [1]. Using the prevalence figure for a drug that treats newly diagnosed metastatic disease nearly triples the starting population and loses credibility immediately.

How do diagnosis rate, line of therapy, and biomarkers shrink the funnel?

Diagnosis rate is the first filter founders skip. CDC estimates 40.1 million Americans have diabetes, but 11.0 million are undiagnosed, about 27.6 percent of adults with the disease [6]. A model that counts total prevalence rather than diagnosed prevalence overstates the reachable population by that margin.

Line of therapy is the second filter, and the losses compound. In a US real-world study of 2,107 patients with non-driver metastatic non-small cell lung cancer, mortality during each of the first four lines ranged from 37.4 to 42.2 percent, and attrition was approximately 60 percent at each line [2]. A third-line drug in that setting addresses roughly one sixth of the first-line population.

Biomarker eligibility is the third filter, and it cuts both ways. Targeted drugs work only in patients whose tumors carry the matching biomarker, and not every patient is tested or has a match [7]. The compensation is in risk: programs using biomarkers for patient selection succeeded from Phase 1 to approval 10.3 percent of the time, versus 5.5 percent without them [4].

What price should I use, and what happens to it after launch?

Launch prices have risen sharply. Across 548 drugs launched from 2008 to 2021, the median annual launch price rose from $2,115 to $180,007, and 47 percent of drugs launched in 2020 and 2021 were priced at or above $150,000 per year [8]. Those are list prices, and list price is not what a company collects.

Net price is what matters, and the gap keeps widening. From 2007 to 2018, list prices for branded drugs rose 159 percent while net prices rose 60 percent, and growing discounts offset about 62 percent of list price increases. Discounts to non-Medicaid payers grew from 23 to 51 percent [3]. I model a gross-to-net discount from launch and widen it over time.

Duration and adherence complete the price step. A $150,000 annual price for a drug taken for eight months yields $100,000 per patient before discounts. For oral cancer therapies, published adherence ranged from 46 to 100 percent depending on population and measurement method [9]. I take duration from the comparable trial and discount it for persistence.

What does a worked TAM funnel look like?

The table below is a hypothetical US solid tumor indication for a second-line targeted therapy. The numbers are illustrative. A real model replaces each assumption with a cited source or a labeled estimate.

At a $150,000 list price with a 25 percent gross-to-net discount, net price is $112,500 per year. With eight months of average therapy, revenue per treated patient is about $75,000. Across 672 patients, peak US revenue is roughly $50 million per year. A 60,000-patient indication became a $50 million drug.

That result is a reason to ask which steps can move, not to abandon the program. An earlier line of therapy, a broader biomarker definition supported by biology, better testing rates, or combination with a standard backbone each change one factor. Built this way, it becomes a strategy document.

How do I sanity-check the number before an investor does?

The top-down check compares the bottom-up result against what analogous drugs earn. If the model says a second-line drug in a mid-sized tumor will earn $2 billion per year and no comparable product has, the model is wrong somewhere. IQVIA reports global cancer medicine spending of $252 billion at list prices in 2024 [10]. That bounds what one product can plausibly capture.

Probability of success is the second check. In an analysis of trials from 2000 to 2015, 13.8 percent of programs entering Phase 1 reached approval, and only 3.4 percent in oncology [4]. Investors apply these factors to reach a risk-adjusted value. Presenting an unadjusted TAM as value invites the correction, so I present both.

I rebuilt a model of this kind for a first-in-class biologic in a rare pediatric leukemia. The initial US addressable market ceiling was roughly $200 million per year. By mapping target biology to significantly larger disease populations and demonstrating efficacy in relevant nonclinical models, the potential US addressable market moved into the multi-billions. That change was real, and it was conditional. Market potential is not revenue.

Investors pushed back in three places. They asked whether nonclinical data justified counting adjacent populations, so I separated the anchor indication from expansion indications and risk-weighted each. They challenged diagnosis and testing rates in the larger populations, so I cited each one. They asked for net price and treated share, not list price and prevalence.

The limits deserve stating. Registry data lag by two to three years, testing and treatment rates vary by region and payer, and a launch price for a drug years from market is a guess. A bottom-up model does not remove that uncertainty. It makes each assumption visible, so disagreements become specific.

This article is part of the Market Positioning Solutions practice at Hamilton Atlas Group — see the full capability list there.

References

1. National Cancer Institute, Surveillance, Epidemiology, and End Results Program. Cancer Stat Facts: Lung and Bronchus Cancer. NCI SEER, 2026.
2. Schoenfeld AJ, Hu C, Rajaram R, et al. Real-world treatment patterns and attrition for non-driver mutation metastatic non-small cell lung cancer in the US. Future Oncology. 2025;21(30):3999-4010.
3. Hernandez I, San-Juan-Rodriguez A, Good CB, Gellad WF. Changes in List Prices, Net Prices, and Discounts for Branded Drugs in the US, 2007-2018. JAMA. 2020;323(9):854-862.
4. Wong CH, Siah KW, Lo AW. Estimation of clinical trial success rates and related parameters. Biostatistics. 2019;20(2):273-286.
5. National Cancer Institute, Surveillance, Epidemiology, and End Results Program. Cancer Prevalence. NCI SEER, 2025.
6. Centers for Disease Control and Prevention. National Diabetes Statistics Report. CDC, 2026.
7. National Cancer Institute. Biomarker Testing for Cancer Treatment. NCI, 2024.
8. Rome BN, Egilman AC, Kesselheim AS. Trends in Prescription Drug Launch Prices, 2008-2021. JAMA. 2022;327(21):2145-2147.
9. Greer JA, Amoyal N, Nisotel L, et al. A Systematic Review of Adherence to Oral Antineoplastic Therapies. The Oncologist. 2016;21(3):354-376.
10. IQVIA Institute for Human Data Science. Global Oncology Trends 2025. IQVIA, 2025.