Spend enough time at the boundary between the laboratory and the market, and a pattern becomes hard to unsee. The programmes that fail are not, for the most part, the ones with weak science. They are the ones where excellent science ran into a gap no one had named — and by the time the gap was visible, it was expensive.
Translational failure tends to concentrate in four places. None of them is glamorous. All of them are addressable, if they are found early.
1. Evidence that doesn’t travel
A result that convinces the people who generated it is not the same as a result that convinces a regulator, an investor, or an acquirer. A mechanism that is inferred rather than demonstrated, a structure–activity relationship that hasn’t been optimized, a dose–response that was never pinned down — each is survivable in the lab and fatal under diligence.
The fix is unromantic: decide, early, exactly what the programme must prove and in what order, then build the evidence to that standard rather than to the standard of internal conviction.
The question is never “is the science good?” It is “will this specific evidence persuade the specific people who hold the next gate?”
2. Regulatory mismatch
The same innovation can be a device in one framing, a drug in another, a cosmetic or a supplement in a third — and the classification determines years of work. Multiply that by the fact that FDA, EMA, ANVISA, PMDA, CDSCO and SAHPRA do not agree on the details, and a programme that assumed one pathway can discover, late, that its lead market requires another.
Mapping the regulatory route for each target market before the evidence plan is locked is not bureaucratic caution. It is what keeps the evidence you generate from being the wrong evidence.
3. Scale-up that breaks
Bench reproducibility is a promise the process makes and often cannot keep. Nanomaterials and complex formulations are especially prone to this: a synthesis that works beautifully at milligram scale behaves differently at kilogram scale, and batch-to-batch variation that was invisible becomes the whole problem. A product that cannot be made consistently cannot be registered or sold, however elegant the chemistry.
The discipline here is to treat manufacturability as a first-class question from the start — characterizing the process, defining specifications, and knowing the route to scale before it is urgent.
4. Commercial positioning that was assumed, not tested
“The market wants this” is a hypothesis, not a plan. Without a defined target product profile — what the product must do, for whom, at what price, against which alternative — a programme optimizes for the wrong attributes and arrives at launch differentiated on dimensions no buyer cares about.
The common thread: decide at gates
What unites these four gaps is timing. Each is cheap to close early and ruinous to close late. That is the entire logic of a go/no-go discipline: break the path into stages, and end each stage with an explicit decision backed by evidence — proceed, pivot, or stop — rather than drifting forward on momentum.
A programme run this way spends money only after the evidence has earned the right to spend it. It is less exciting than betting the whole thesis at once. It is also how science actually reaches the market.
- Discover — separate what is proven from what is assumed.
- De-risk — audit against readiness criteria and close the critical gaps.
- Position — lock the regulatory route and the target product profile.
- Launch — coordinate partners and filings toward scalable market entry.
The gaps are predictable. That is the good news. A programme that goes looking for them — before diligence, before scale-up, before launch — is a programme that gets to keep its upside.