Why Good Science Still Fails to Drive Decisions in Biotech

A biotech team can spend years building a credible body of evidence and still leave an important meeting without a clear decision. The studies were well designed. The data hold up. Every technical question raised in the room gets answered competently. And the meeting still ends with some version of “we need to review this internally.”

I have watched this happen for more than a decade, and it shows up most consistently in cross-functional meetings. It has little to do with seniority, credentials, or how experienced the people in the room are. The real issue is that these meetings tend to look like a series of departmental results presentations, when what the room actually needs is a clear answer on whether and how to move forward.

Sometimes more evidence really is required: another analysis, longer follow-up, or a better-controlled study. But not always. Often the evidence already in the room is relevant to the decision at hand. What is missing is not more data. It is a visible line connecting what was found, what is still uncertain, and what the organization is actually being asked to do next.

Scientific rigor and decision clarity sit close to each other, but they are not the same skill. Science exists to establish what was observed, how it was measured, and how confidently the results can be interpreted. A development, investment, partnering, or resource-allocation decision comes at the same information from a different angle: What does this evidence change? Which uncertainty actually matters right now? What is already well supported today? What should happen next? When those four questions stay unanswered, even strong science struggles to move a program, a partnership, or an organization forward.

A Scientific Story and a Decision Story Answer Different Questions

Technical teams tend to organize information the way the work happened. Biological background first, then study design, methodology, findings, and a scientific conclusion. That order makes sense to the people who did the work because it follows the investigation itself.

But the person responsible for the decision usually is not trying to reconstruct that whole process. They are trying to figure out whether the evidence in front of them supports a specific course of action.

Take a translational team presenting target-expression data, assay performance, model selection, subgroup findings, and a handful of biological hypotheses. Every part of that presentation can be scientifically sound. The actual decision on the table, though, might be much narrower: is the biomarker strategy mature enough to shape the design of the next clinical study? If the deck does not help the room answer that one question, the content is informative, but it is not decision-ready yet.

This gap shows up everywhere in biotech development, and it does not look the same twice. A research team might demonstrate real biological activity without clarifying whether that is enough to justify candidate selection. A clinical team might show encouraging early activity without stating whether it changes the dose strategy, the patient segment, or the design of what comes next. CMC updates often describe genuine process progress while leaving the consequences for supply, comparability, timeline, and capital unspoken. Three functions later, someone finally asks, “Wait, does this push the IND date?” A BD summary can lay out an asset thoroughly without answering the question a partner’s team is actually there to ask: which piece of this evidence matters most given their portfolio and tolerance for development risk?

In every one of these cases, the science holds up fine. The communication still has a second job it has not finished doing.

Scientific story

1. 01Background
2. 02Methods
3. 03Findings
4. 04Conclusion

→

Translation gap

1. Relevance
2. Interpretation
3. Uncertainty

→

Decision story

1. 01What changes?
2. 02What remains open?
3. 03What should happen next?

Why Strong Evidence Still Fails to Move the Conversation

When a presentation fails to produce a decision, the instinctive response is almost always to add more. I have watched this happen even with some of the most scientifically rigorous people in the room. I once worked for a manager with a strong technical background who, whenever he felt a department’s summary had not given him enough to make a call, would ask for more supporting evidence. One more analysis. A bit more background. A supporting experiment pulled out of the appendix and into the main deck. A few slides added in anticipation of questions that might come up.

What started as a thirty-page deck would quietly grow to sixty. By the time it reached the board, most of the room could no longer follow the argument, and the decision that was supposed to happen that day got pushed back again.

The deck got longer. It did not get clearer.

That is because the actual problem is usually misdiagnosed. The audience is not necessarily asking for more evidence. They are struggling to understand how the evidence already on the table should shape what happens next.

The Story Follows the Work, Not the Decision

Biotech programs generate information across functions, over years. A target gets selected. Models get developed. Assays get qualified. Candidates get compared. A dose gets chosen. Clinical data start to come in, and regulatory, manufacturing, IP, and commercial questions develop right alongside the science. Naturally, teams present in that same order because it is how they lived it.

The problem is that the chronology of the work is not always the logic of the decision. A leadership team deciding whether to keep funding a program does not need to revisit every experiment that got the program to where it is. It needs to know whether a small number of conditions have been met: Is the biological hypothesis still credible? Does the current evidence support a viable path forward? Can the next round of investment resolve something that would materially change the program’s value? Those questions usually draw on research, translational science, clinical, CMC, and commercial input all at once. If each function presents its own update separately, the room is left to assemble the conclusion by itself, in real time, out loud.

This is especially common in cross-functional reviews. Each team gives a technically correct update. The meeting still produces no shared answer. One group sees progress. Another sees unresolved risk. A third has identified a dependency nobody else in the room has connected yet. The presentation documents activity. It does not make the relationship between these views visible.

A dose-selection discussion is a good example. It might involve exposure across dose levels, target engagement, pharmacodynamic response, observed activity, safety findings, assay limitations, and the proposed design of the next study. Presented one function at a time, these are seven separate updates. Put together deliberately, they become a single dose rationale. One format reports what each team knows. The other helps the organization decide what the combined evidence actually supports.

Material Uncertainty Is Present, but Rarely Structured

Uncertainty is almost never absent from a biotech presentation. It is simply scattered, and often invisible to the person presenting it. A study limitation sits in small print under a chart. A missing analysis gets mentioned once, verbally, and never returns. An assay issue lives in a technical appendix nobody opens during the meeting. A regulatory dependency shows up on a timeline three sections later. Each point was technically disclosed. The audience can still miss what it means.

The real question is not whether uncertainty was mentioned somewhere. It is whether the communication makes clear which uncertainty could actually change the conclusion.

Not every limitation deserves the same weight. A small sample size might be central to interpreting an early efficacy signal, but almost irrelevant to whether target engagement was observed at all. Limited follow-up can matter enormously for durability or safety claims while barely affecting whether biological activity was detected. An assay that is perfectly acceptable for exploratory work may be nowhere near good enough for prospective patient selection. A manufacturing step that is routine at one stage can become a gating item before the next clinical milestone, and the team that already knows this rarely says it out loud because to them it is obvious.

A useful presentation separates general incompleteness from decision-critical uncertainty. That distinction lets the audience move past the vague “what else do we need to know?” and get to something more useful: which unresolved issue is most likely to change the current conclusion?

One way to phrase this directly: the current evidence supports [conclusion], while [specific uncertainty] remains unresolved because [why it matters]. The next decision depends on [analysis, study, review, or milestone].

That is more useful than “additional validation is required,” because it tells the reader what is already supportable, where the evidence boundary actually sits, and what needs to happen before the conclusion can get stronger.

The Next Action Often Stays Implicit

Most scientific presentations end with a summary of findings. That is fine for a conference talk. It is usually not enough for a decision meeting.

A scientific conclusion answers: what do the current findings indicate? A decision conclusion has to answer a second question too: what should the organization actually do with that interpretation? The team that prepared the material usually knows the answer already. It is rarely obvious to everyone else in the room, and asking them to guess is how “let’s take this offline” happens.

Picture a program review where the team presents target engagement, early activity in a selected subgroup, variable exposure, and a safety dataset that is still immature. The scientific conclusion might read: the findings support continued evaluation of the mechanism in the selected population. Reasonable enough, but it does not define a decision. The organization could continue the current dose-escalation plan, revise the dose strategy, expand the subgroup, complete an exposure-response analysis first, wait for more safety follow-up, or stop the program because the next step is not sufficiently differentiated from where things already stand. The same evidence can justify very different actions depending on the organization’s objectives, risk appetite, capital position, and the alternatives it is weighing that program against.

That is why the proposed next step should be stated outright rather than left for the room to infer. A useful decision conclusion makes four things visible: the proposed action, the evidence behind it, what is still open, and when the decision will be revisited. For example: proceed with the planned cohort expansion while completing the exposure-response analysis. Current activity and target-engagement findings support continued evaluation, but final dose selection should stay open until the updated analysis and additional safety follow-up are reviewed at the next development committee meeting.

This does not pretend the evidence is complete. It just shows how the organization can act despite the gaps, which, in practice, is most of what a development team is actually paid to do.

What Decision-Ready Communication Makes Visible

Decision-ready communication is not a special slide template. It is a way of organizing complex information around a defined purpose, and it usually comes down to four things being visible at once.

1. 1 The decision being considered

   Define the specific decision that needs to be made now.

2. 2 The evidence that bears on it

   Present the evidence that materially informs the decision.

3. 3 The uncertainty that could change it

   Name the material uncertainty that could change the current conclusion.

4. 4 The next action and decision checkpoint

   State the proposed next step and when the decision will be revisited.

The decision being considered. “Review the program” is not a decision. Neither is “discuss the data.” The decision needs to be stated in operational terms: should the company advance the candidate into IND-enabling studies? Is the evidence sufficient to support initial partner diligence? Should the next trial focus on a defined biomarker-positive population? Does the program warrant the next tranche of capital? A clearly stated decision draws a boundary around the story and helps the team separate what is essential from what is merely available.

The evidence that actually bears on that decision. Not everything belongs in the core narrative, and the most scientifically interesting result is not always the most decision-relevant one. What matters is the evidence that could materially change the current judgment: human biology supporting the target hypothesis, a design limitation affecting interpretation, exposure data influencing dose selection, a safety finding altering the development path, or a manufacturing constraint changing timeline and capital needs. Everything else can still live somewhere: an appendix, a technical review, a data room, or a separate functional discussion. Taking it out of the core story does not diminish its scientific value. It stops material meant for a different audience from burying the decision logic meant for this one.

The uncertainty that could actually change the conclusion. A useful presentation distinguishes between what has been directly observed, what is supported but not yet confirmed, and what is still a hypothesis or future objective. Blurring these categories is how fragile stories get built. A planned product attribute should not be written as an achieved result. An exploratory subgroup observation should not be presented as confirmatory evidence. Internal expectations should not be described as regulatory alignment. The audience needs to see the current conclusion and its actual boundary, not to make the program look weaker, but to make the team’s reasoning something other people can inspect and challenge.

The next action and the checkpoint attached to it. A recommendation should say more than what the team plans to do next. It should explain what that step is meant to clarify and how the resulting evidence will feed into a later decision. A clinical study is an activity. A safety, PK, and PD dataset is an output. Evidence is what gets learned from that output. A milestone is the development point reached. The decision is what actually changes because new evidence now exists. Collapse those layers into each other and teams start mistaking activity for progress. Completing the study feels like an achievement even when nobody can say what it proved.

The Reader Should Not Have to Build the Argument

Technical experts see connections that are invisible to almost everyone else in the room. They know which assay is actually reliable. They remember why a particular model was chosen two years ago. They know which result changed their own confidence in the program. Much of this context never makes it into the deck because it feels too obvious to the people who lived it to bother writing down.

The audience does not share that mental map. An investor may see promising data without understanding what the next financing round is meant to establish. A partner may find the asset credible while remaining unsure which diligence question should be raised first. An executive may follow the science perfectly well and still not see how the program stacks up against everything else competing for the same capital and attention.

The presentation’s job is to make those connections something the reader can examine. The decision, the evidence bearing on it, the limits of that evidence, the team’s interpretation, and the action attached to it should all be visible. When they are, disagreement gets more useful. People can push back on the evidence, the interpretation, the risk read, or the proposed action directly, instead of asking vaguely for more information because they cannot tell what to challenge.

Clarity Does Not Require Artificial Certainty

Biotech teams sometimes avoid taking a clear position because the evidence is not complete yet, worried that a firm recommendation will look overconfident. What comes out instead is language that is technically careful but operationally useless: the data are encouraging, and additional work will be required to further evaluate the opportunity.

Nobody can really disagree with that sentence. Nobody can act on it either.

A more useful version stays cautious and still takes a position: the current evidence is sufficient to continue the program through the planned validation study, but not yet sufficient to support broader patient selection. The next decision should follow prospective biomarker performance and updated clinical follow-up. That sentence commits to something and marks its own boundary at the same time.

The same discipline applies to smaller claims. “Best-in-class efficacy,” “clean safety profile,” and “validated biomarker” all sound persuasive on a slide, but they usually force the reader to spend the first few minutes figuring out what the words actually mean before evaluating anything. Precision tends to be more persuasive than confidence. State the population, the study design, the analysis set, the finding, the timepoint, and the data maturity. Name the material limitation. Then say what the result actually supports. A preliminary result can still justify the next experiment. A limited dataset can still be strategically important. An unresolved risk can still be manageable. The reader just needs to see why, in that order, before believing it.

Good Science Creates the Opportunity. Communication Helps the Organization Act on It.

Biotech decisions are almost never made with complete information. Programs move forward through a sequence of judgments about how strong the current evidence is, how much the remaining uncertainty matters, and how valuable the next step would be.

A presentation cannot remove development risk. It cannot replace scientific review, cross-functional expertise, or due diligence, and it should not try to tell an investor, executive, or partner what to decide. What it can do is make the decision structure visible, so the evidence becomes easier to evaluate, the uncertainty becomes easier to argue about honestly, and different functions can finally see how their piece connects to everyone else’s.

The science is still the foundation. It just does not move through an organization on evidence alone. It moves through shared interpretation, explicit choices, and a clear sense of what the next step is supposed to resolve. That is the difference between presenting scientific information and building communication that can carry a decision.

Structure the science. Move the conversation.

NarraFlect develops decision-ready communication systems for biotech teams working across fundraising, partnering, scientific communication, and cross-functional decision-making.