Drug Development Phases Explained: A Field Guide for Founders and Generalist Investors

Why does drug development have phases at all?

The phase structure exists because you cannot answer every question about a new medicine at once, and the questions are ordered by how much they cost to answer and how much harm a wrong answer could do. Each phase is a gate designed to retire one class of risk before you spend the money required to confront the next. You establish that a molecule does something biologically interesting before you make it at scale, you establish that it is not overtly toxic in animals before you give it to a human, and you establish that it is tolerable in a small number of humans before you test whether it treats disease in many.

For a founder or a generalist investor, the single most useful reframing is this: a clinical phase is not a milestone the company passes, it is an experiment the company runs. The value of the asset does not come from being in Phase 2. It comes from the readout of the Phase 2 experiment. That distinction governs almost everything about how these companies are financed and how their worth changes over time.

What happens in discovery and preclinical work?

Discovery is where a program identifies a biological target implicated in disease and finds a molecule, an antibody, a cell therapy, or another modality that engages it. The scientific claim being tested here is one of mechanism: if you modulate this target, you should change the course of this disease. Much of the highest-quality translational science originates in academic labs, which is one reason university spinouts are so central to early-stage life-sciences venture.

Preclinical development takes a promising candidate and asks whether it is safe enough, and manufacturable enough, to justify human testing. This work covers pharmacology, toxicology, formulation, and early chemistry, manufacturing, and controls, commonly abbreviated CMC. Toxicology is typically conducted in more than one animal species. The risk being retired is catastrophic toxicity and undevelopability. Preclinical work is generally cheaper and faster than clinical work, which is precisely why it sits before it. A generalist should note that a large share of programs die here, quietly, before a single patient is ever involved.

What is an IND, and why does it matter so much for early investors?

Before a drug can be given to humans in the United States, the sponsor files an Investigational New Drug application, or IND, with the Food and Drug Administration. The IND packages the preclinical safety data, the manufacturing information, and the proposed clinical protocol into a single submission. Under the standard process the proposed study may proceed after a defined review window, commonly cited as roughly thirty days from the FDA’s receipt of the IND, unless the agency places the program on clinical hold.

IND-enabling work is often a decisive value inflection point for pre-seed and seed investors, and it is worth stating plainly. Getting a program from a compelling academic result to a cleared IND can be one of the most capital-efficient value-creation steps in the entire lifecycle. It converts an interesting paper into a bona fide clinical-stage asset that later, better-capitalized investors will typically pay a step-change premium to own. A great deal of what early venture actually funds is the unglamorous, expensive, de-risking work that stands between a publication and an IND.

What do Phase 1, Phase 2, and Phase 3 each de-risk?

The three phases of human trials are frequently misunderstood as three attempts at the same question. They are not. Each asks something different.

Phase 1 primarily tests safety and tolerability, usually in a small group, often healthy volunteers, though in oncology and other serious conditions it is typically run in patients. It establishes an initial dose range and the pharmacokinetic profile, meaning how the body absorbs, distributes, and clears the drug. Phase 1 retires the risk that the drug is dangerous to people at a plausible dose. It does not, by design, prove the drug works.

Phase 2 is where the central bet of most programs is finally tested: whether the drug produces a therapeutic effect in patients who have the target disease. This is the first real efficacy signal, and it is consistently one of the hardest gates in the pipeline. Industry data has long shown Phase 2 to be the phase the largest share of programs fail to clear, because this is the moment biology is asked to deliver on the mechanistic promise made back in discovery. A generalist investor who remembers only one attrition fact should remember that Phase 2 is where the graveyard is.

Phase 3 confirms efficacy and safety in a larger, more representative patient population, typically through well-controlled, often randomized trials sized to satisfy regulators. It retires the risk that the Phase 2 signal was a statistical fluke or an artifact of a small, favorable population. Phase 3 is generally the most expensive stage, routinely accounting for the largest share of clinical trial spend, which is why programs that reach it are often financed by crossover and public-market investors rather than seed funds.

Why is attrition so high, and what do the numbers actually say?

The honest headline is that most drugs fail, and they fail expensively. Industry analyses have long suggested that the likelihood of a program entering Phase 1 eventually reaching approval sits in the single digits. These figures vary meaningfully by source and by methodology, so they should be read as approximations of a hard truth rather than precise constants, but the direction is not in dispute.

The variation by therapeutic area is large and matters more than the blended average. Oncology has historically shown some of the lowest cumulative success rates from Phase 1, because cancer biology is heterogeneous and endpoints are demanding. Other areas, including some vaccines and certain well-understood metabolic or hematologic indications, have shown materially higher rates. A blended industry number is close to useless for evaluating a specific asset; the relevant base rate is the one for that modality, that target class, and that indication.

Attrition is high because the phases are, correctly, adversarial toward the hypothesis. The whole point of the structure is to kill wrong ideas as early and as cheaply as possible. From a portfolio perspective this is a feature, not a bug: the discipline that makes any single program likely to fail is the same discipline that makes the survivors genuinely de-risked.

What are the NDA and BLA, and what happens at approval?

Once Phase 3 succeeds, the sponsor assembles the entire body of evidence into a marketing application. For a small-molecule drug this is a New Drug Application, or NDA. For a biologic, such as an antibody or a cell or gene therapy, it is a Biologics License Application, or BLA. The FDA then reviews the dossier against defined target timelines, commonly cited as roughly ten months for a standard review and six months for a priority review, measured from the filing date.

The relevant fact for investors is that programs reaching this stage convert to approval at a comparatively high rate, since most of the scientific risk has already been retired by the time an application is filed. What remains is largely a question of data completeness, manufacturing readiness, and regulatory judgment. This is why the value curve of a successful asset tends to be steep across late clinical development and comparatively flatter afterward.

What is post-market surveillance, and why does it exist?

Approval is not the end of evidence generation. Phase 4, or post-market surveillance, monitors a drug once it is used by a far larger and more varied population than any trial can enroll. Rare adverse events, long-term effects, and interactions that could not surface in a controlled trial of limited size and duration can become detectable at real-world scale. Regulators can require post-marketing studies as a condition of approval, and can restrict or withdraw a drug if the real-world safety picture deteriorates.

For a company, Phase 4 is also where commercial reality is tested: adoption, reimbursement, and durability of the clinical benefit in ordinary practice. These are business risks layered on top of the scientific ones, and they are worth naming because a technically approved drug is not automatically a successful product.

How does a pre-seed or seed check fit before Phase 1?

Pre-seed and seed capital in life sciences almost always does its work upstream of the first clinical trial. The money funds the translational bridge: taking a validated academic finding through the preclinical, manufacturing, and regulatory work required to reach a cleared IND, and sometimes into an early Phase 1. This is deliberately the stage before the largest costs and the most-watched readouts, which is one reason entry valuations tend to be lower and the potential step-up largest.

The financing rhythm of the sector follows the phase structure directly. Value tends to inflect in steps at defined milestones, IND clearance, a clean Phase 1 safety and dose readout, and above all a positive Phase 2 proof of concept, and rounds are typically raised to reach the next such milestone rather than to fund a fixed number of months. A well-constructed early plan is legible as a sequence of value-creating experiments, each with a defined cost, a defined readout, and a defined effect on what the company is worth. Founders who can articulate that sequence are far easier to underwrite than those who present a diffuse roadmap.

The practical implication for founders is to raise against the milestone, not the calendar. The question a disciplined seed investor is really asking is what the next data readout costs, what it will prove, and how much it changes the value of the asset if it succeeds.

What does Sonnerie look for in translational science at the earliest stage?

Sonnerie invests at pre-seed and seed in healthcare and life-sciences companies, frequently university spinouts, at the point where rigorous science is being turned into a clinical asset. At that stage there is no clinical efficacy data to rely on, so the diligence is necessarily about the quality and honesty of the underlying signal.

We look for a mechanistic hypothesis that is specific and falsifiable rather than merely plausible, and for a founding team that already understands which experiment would prove them wrong. We look for a clear, honest line of sight to a cleared IND, with the preclinical, manufacturing, and regulatory work scoped realistically rather than waved past. We look for indication selection that respects the base rates, meaning a first target where the biology gives the program a genuine chance rather than the largest possible market. And we look for founders who treat attrition as information: people who will kill their own program on a bad readout are, paradoxically, among the people most worth backing, because the entire value of the phase structure is its willingness to say no early.

This is what we mean by finding the signal in the noise. In an asset class where most programs fail and much of the failure is knowable in advance, the edge is not optimism. It is the discipline to hear the real breakthroughs early and to fund the specific experiments that separate them from the rest.