Biotech does not fail because biology is "hard" in some abstract sense. It fails in very specific, depressingly repeatable ways: unexamined assumptions, performative analysis, and an endless capacity for self-deception dressed up as optimism. Due diligence, done properly, is not a ceremonial rite to justify enthusiasm. It is a disciplined attempt to prove that an asset, a company, or a thesis is guilty until it has genuinely earned acquittal.

Over the last few years, I've watched many diligence processes drift toward comfort theatre. Beautiful decks, well-curated expert calls, checklists ticked off with no real attempt to break the story. Meanwhile, the macro environment has become far less forgiving. Capital is more selective. Biology is more complex. Regulators are less indulgent of fairy tales. In that world, treating diligence as a formality is not just naïve. It's negligent.

What follows is a clear-eyed view of what good diligence is, why most of what passes for diligence falls short, and how to build a culture that is more interested in being right than in being reassured.

1The Problem: Narrative First, Evidence Later

Most bad diligence starts with a story and works backward to find supporting exhibits. The narrative usually sounds something like this:

• The modality is hot, the indication is large, and the market is "underserved."
• The management team is "strong" and "serial," which is often code for "they've raised money before."
• The science is "cutting-edge," a phrase that should trigger immediate suspicion, because tools are rarely the rate-limiting step.

From there, the questions get stacked in the wrong order. People ask whether the trial design is "standard," whether the IP looks "solid," whether the valuation is "reasonable for the peer set." All of this presumes something critical: that the core biological claim is actually true, and that the asset has a path to a meaningful clinical and commercial endpoint.

In a surprising number of cases, that presumption does not survive contact with the data.

A disciplined skeptic does something different. Instead of asking "how could this work," they repeatedly ask "why might this fail" and "what would have to be true in the world, beyond this slide deck, for this to have the impact implied by the valuation." That shift in orientation sounds small. It is not. It flips the burden of proof from reality onto the asset.

2Orientation Before Detail: Where Are We, Really?

The single most underrated step in diligence is orientation. Before you plunge into mechanism, trial design, or TAM calculations, you need to answer a simpler question: where, exactly, are we on the curve?

The answer matters more than most people admit. A first-in-class asset in a truly unmet, biology-driven space carries different risks than a tenth PD-1 in a jaded payer environment, even if both come packaged in glossy immunology language. The former is primarily a scientific and translational risk. The latter is primarily a commercial and access risk.

Orientation also forces you to confront capital cycles. We are not in the world where "novel target, solid mouse data, credible PI" guaranteed a cheque. The valley of death has moved downstream, syndicates have grown more selective, and big pharma has more ways to source external innovation than it once did. That reality should be part of any diligence conversation. Whatever the specifics of the current cycle, if your investment thesis relies on a capital environment more forgiving than the one actually available, you are not doing diligence. You are writing fan fiction.

3Skepticism as an Operating Stance, Not a Personality Quirk

Skepticism is often treated as a personality trait, like pessimism or optimism. In this industry, that framing is wrong. Skepticism is an operating stance. It is the tool that lets you look at a beautifully constructed narrative and say, "Fine. Let's assume you are wrong. What breaks first?"

There are at least three buckets where skepticism earns its keep.

a) Clinical data and trial design

Early oncology and immunology programs are experts at generating seductive signals that evaporate in Phase 3. We have literally decades of experience with cancer vaccines that work in mice, glimmer in Phase 1, and die in large controlled trials. Yet every new wave comes with the same breathless language: "positioned for success", "multiple reasons for optimism", "this time is different." The base rates justify the caution: across more than 21,000 compounds, the overall probability of a drug that enters clinical trials going on to approval is roughly one in seven, and in oncology it is closer to one in thirty.¹

A skeptical diligence team does not ban those phrases. It asks, calmly:

• What is the mechanism by which this program avoids the last three generations of failure modes in this field?
• How sensitive are the endpoints to functional benefit rather than just biomarker movement?
• Where exactly is the tail in this Kaplan–Meier curve, and is the valuation anchored on the mean or the outlier?

In sarcopenia, for example, you don't get to claim victory because DEXA scans show more lean mass. If gait speed, chair rise, grip strength, falls, and institutionalization rates don't move, payers will treat your "success" as ornamental. The skeptic cares about usable muscle, not pretty scans.

b) Mechanistic claims and translational depth

Biotech has an almost inexhaustible supply of clever mechanisms that look compelling at the level of a Nature figure and collapse in contact with heterogeneous human disease. Fibrosis, cancer vaccines, senolytics, microbiome therapeutics, immunometabolism; pick your favorite graveyard.

Skeptical diligence demands translational depth. That means asking:

• Is there a credible chain from the model system to the real clinical population, with acknowledged breaks and uncertainties?
• Have we honestly accounted for age, comorbidity, prior treatment, and microenvironmental factors that have killed similar programs?
• Are we extrapolating from a cartoon schematic of the pathway, or from multiple independent lines of evidence that converge on the same biological point?

True orthogonal validation is rare. Most programs have something less robust, and diligence should price confidence accordingly rather than pretending every target arrived with celestial endorsement.

c) Macro narratives and capital expectations

The industry tends to oscillate between two equally lazy stories. One insists we are entering a golden age where AI-native R&D and cardiometabolic blockbusters will restore growth and margins almost by default. The other asserts that the business model is structurally broken, condemned to regulated-utility returns by pricing reform and payer aggression.

Skepticism cuts through both. It accepts that AI materially improves certain tasks while refusing the leap to systemically de-risked pipelines. It acknowledges headwinds without assuming that every new modality will slide into commodity land on the same timeline as small molecules.

In diligence, that translates to a simple discipline: every asset should be evaluated in the context of both its scientific opportunity and the actual funding architecture it will face over the next decade. If your forecast assumes a fantasy TSR for biotech relative to the broader market, you are not being skeptical. You are simply hoping to be bailed out by valuation multiple uplifts.

4AI, Data Governance, and Shadow Risk in Diligence

One of the more novel failure modes creeping into diligence now is AI-related. Early biotechs are using AI everywhere: Copilot for documentation, meeting bots for transcription, CRO dashboards, LLMs for slide drafting. Many cannot explain how, or where, the data flows through these systems.

The risk here is not a Hollywood scenario where a rogue model steals your IP. It's quieter and more insidious: unpublished sequences, draft patents, BD decks, and confidential clinical datasets leaking into proprietary training pipelines or "helpful" tools whose configuration nobody has audited. The first time this shows up meaningfully is not a cyber attack. It is an awkward line of questioning in a diligence session and a silent haircut on valuation.

Good diligence now includes AI governance as a first-class topic. That means:

• Mapping which systems touch sensitive data, and what their default retention and training settings are.
• Asking who owns the configuration and whether there is a coherent policy, not just a slide about "trusting our vendors."
• Treating AI outputs as junior analyst work that requires human review, not as unexamined ground truth.

Used well, AI is an excellent accelerator for R&D. Used badly, it quietly undermines the integrity of the dataset you are trying to generate. A rational investor should care about both.

5Value Lives in Arithmetic

For all the lofty talk about vision and transformative science, biopharma value ultimately lives in cash flows. That doesn't mean every diligence conversation needs a 300-row spreadsheet. It does mean that qualitative enthusiasm should be tied back to a risk-adjusted net present value the team can actually defend.

Risk-adjusted NPV forces three useful disciplines:

• You must be explicit about probabilities of success at each stage, rather than hand-waving about "de-risked assets."
• You must tie peak sales and pricing assumptions to actual competitive and access landscapes, not just analogue slides.
• You must confront the impact of tax, clawbacks, rebates, and fiscal drag instead of quietly assuming net prices track list prices.

When you do this honestly, a surprising number of seemingly attractive programs reveal themselves as value-destroying at the portfolio level. The failure is rarely the headline probability of success. It is usually one input that qualitative discussion had waved past: a net price that collapses once rebates and access restrictions are applied, a peak share that assumes the incumbent stands still, or an endpoint the market will not actually pay for. Change that single input and an asset that looked compelling on a slide turns value-negative on a model. That is not an argument against funding innovation. It is an argument against pretending that every shiny mechanism deserves a premium.

Risk-adjusted net present value is not elegant, and it is certainly not infallible, but it imposes useful discipline. It forces explicit assumptions about technical success probabilities, timelines, pricing, access, peak penetration, geographic mix, and the erosion caused by rebates, taxes, clawbacks, and competitive response. That is valuable precisely because it prevents qualitative enthusiasm from floating free of economic consequence.

Risk-adjusted models are not perfect. They rely on inputs that can be wrong. But they force teams to put their assumptions on the table and argue about them in daylight. That alone makes them superior to the more common approach of "the valuation feels about right for the space."

6Culture: Making It Safe to Kill Bad Ideas

The hardest part of disciplined diligence is not the analysis. It is the culture. Death, taxes, and stupidity are the three inevitabilities in this business. You cannot repeal any of them, but you can decide how much stupidity you tolerate inside your own decision process.

Most organizations pay lip service to "learning from failure" while quietly punishing people who pull the plug on popular projects. That is how you end up with programs that should never have left in vitro, acquisitions done for vibes, underpowered pivotal trials, and go-to-market plans that ignore how clinicians actually behave. These are not tragedies of uncertainty. They are preventable own goals.

A serious diligence culture does a few unglamorous things:

• It runs ruthless pre-mortems before green-lighting a program or deal: assume it failed, then enumerate the most plausible reasons and what evidence you would need now to refute them.
• It makes "I don't know" an acceptable sentence in senior meetings, rather than a career-limiting confession.
• It builds incentive structures that reward killing bad ideas early instead of dragging them to a Phase 3 funeral.

This is where skepticism stops being an analytical stance and becomes an organizational advantage. Teams that can admit uncertainty, dissect their own biases, and walk away from attractive but fragile stories tend to compound value quietly over time. Teams that cannot mostly produce good conference slides and disappointing TSR.

7The Consultant's Role: Less Theatre, More Orientation

Consultants in this space have a choice. We can sell reassurance, or we can sell clarity. The market has an oversupply of reassurance. There are plenty of beautifully formatted "diligence reports" that repeat the company's narrative in more polished language, add three expert quotes, and conclude that "while uncertainties remain, the opportunity is compelling."

Clarity is harder. It often sounds like:

• "The science is interesting but unlikely to translate at the effect size implied by this valuation."
• "The capital stack you are relying on has already moved; by the time this program hits its next inflection, the funding window you are assuming will probably be closed."
• "Your biggest risk is not the biology; it is your own governance. You do not currently have the culture to kill this if the next dataset looks like X instead of Y."

Clarity does not always make you popular. It does, however, make it more likely that your clients allocate their scarce capital to assets that have a realistic chance of turning mechanism, data, and execution into actual NPV.

In that sense, due diligence is less about predicting the future and more about protecting your downside from your own enthusiasm. You won't eliminate death or taxes. Biology will continue to surprise you. Regulators will continue to tighten and relax at unhelpful times. The one variable you can control is how rigorously you interrogate your own stories before you bet real money on them.

That is what "guilty until proven innocent" actually means. It is not cynicism. It is simply the minimum standard of care in a sector where hope is cheap and capital is not.

Author

Anthony Walker, PhD

Anthony draws on more than 35 years of experience, including over a decade spent building and managing a biotechnology company and over 20 years as a management consultant to the pharmaceutical and biotech industries.