The Milestone Was Real. The Decision Was Still Open

The antibody program had completed Phase I smoothly. It still did not move into Phase II. Several product lines were competing for the same resources, and the company was uncertain about the eventual market response if this treatment for viral infection advanced through Phase III and toward launch. The clinical milestone was real. The next allocation decision was still open.

I worked on that program, and my judgment was that the portfolio and market questions should have been examined when the project was initiated. Earlier assessment would not have guaranteed a different outcome. It would not have removed commercial uncertainty. But it could have made the assumptions behind starting and continuing the program visible before development had progressed so far.

Instead, those concerns became consequential after Phase I. Investors questioned whether the drug had been assessed carefully at the outset or had been pursued mainly to expand the pipeline. Despite the clinical progress, the company lacked confidence about continuing. The project was delayed for several years.

That experience left me with a straightforward conclusion: completing a difficult stage does not by itself make the next choice easier. The test belongs before the result arrives. Before approving the capital, the team should write the materially different result branches, decide which actions each could support, and check that those choices will still be available when the evidence arrives.

What Phase I Could Not Decide

Nothing in this argument makes Phase I scientifically or operationally unimportant. The program had completed it smoothly. The problem was narrower and more practical: that progress could not settle portfolio priority, future market confidence, and the next commitment of resources.

What stood out to me was not a failure of Phase I. We had asked its completion to carry a portfolio decision that the clinical milestone had never been designed to answer.

This is where value-inflection language can obscure the decision. Expense, technical difficulty, and a place on the development timeline may all make a milestone important. Before the work is funded, however, the team should be able to say what choice its result could change.

In some plans, progress is the honest and sufficient description. Calling it that does not diminish the execution involved; it keeps the claim aligned with what the proposed work can support.

The Test Belongs Before the Result

The practical error is usually made before a dataset or report exists. A plan names a study, a development stage, a completion date, and a budget. Because the work is concrete, the reason for funding it can appear equally concrete even when the later choice remains undefined.

The central question here is not how to present completed evidence more clearly. It is whether capital is being committed to a test whose possible results leave different actions genuinely available. If the action has already been fixed, better evidence communication cannot restore a decision point that the plan removed.

For this article, I reserve value inflection for a point where new evidence can change a consequential choice for a defined decision owner. I use the term as an ex ante decision-design test, not as a technical standard or a promise of valuation effect. What counts as a meaningful change will vary by program. The discipline is to state the intended decision and make the possible differences in action inspectable before capital is committed.

Start with the uncertainty preventing a decision today, then examine whether the proposed work can distinguish among the paths the organization is actually willing to take. An event may sound important and still leave the intended action unchanged.

Some work will be necessary even when no immediate branch exists. Documentation, supply preparation, method development, and operational readiness may all deserve funding because they make later work possible. They do not need inflated value-inflection language to justify their place in the plan. The distinction matters because it prevents a necessary activity from carrying a promise it was never designed to fulfill. The work can be funded honestly now, while the later decision test remains visible.

Write the Counterfactual Branches Before Funding the Work

Start with the next decision in operational terms. Advance, adapt, pause, stop, or choose one program over another are decisions. “Review the data” is an event. It says nothing about what the organization is prepared to do after the review.

Then write down the materially plausible result categories. They do not need to predict every observation or force continuous data into a neat positive-negative split. They need to be different enough that a decision maker could reasonably interpret them differently.

For each result, record the supported interpretation, the uncertainty that remains, and the action it could justify. Name the person or group that owns the choice. This is where vague milestone language is exposed: the work may be well specified while the consequence of learning is not.

Now ask the counterfactual question: if the evidence were materially better, worse, mixed, or inconclusive, would the action change? If every branch continues along the same path, the work may still be required. Its decision function, however, is weak or undefined. The capital is financing execution without yet preserving a discriminating test.

The value of this exercise is not the diagram. It is the disagreement the branches expose. A scientific lead may believe a weaker result still supports another study, while a portfolio owner may see the same result as a reason to redirect capital. That difference should surface before the result arrives, not in the meeting where everyone is already under pressure to interpret it.

A negative result can be valuable when it credibly supports stopping a path, changing the design, or moving resources elsewhere. A positive result can have little decision value when it only confirms what the organization already assumed and leaves the gating uncertainty untouched. What matters is the consequence the evidence can support for the chosen path, not whether the result sounds favorable.

Ambiguity deserves its own branch. An unclear result may expose a limitation in the method, insufficient maturity, or a genuinely unresolved hypothesis. Those explanations require technical judgment; they should not be assigned by habit. The useful question is whether the ambiguity narrows the next problem or changes the next action. Repeating the same work under the same interpretation boundary may produce another output without producing much learning.

Timing can change the answer as well. Evidence may be decision-relevant and still arrive too late, after manufacturing scale-up, hiring, or another commitment has reduced the available options. A sound map records when the decision must be made and whether an interim action can be reversed. The evidence required for a limited, reversible step should not be confused with the evidence required for a full commitment.

The organization must also be willing to accept the branches it writes down. A stop criterion that nobody intends to apply is not a decision boundary. It is a presentation device for a continuation decision that has already been made.

The Manufacturing Decision That Made This Concrete

I encountered the same distinction in manufacturing. Management wanted drugs that were still in preclinical or research stages to be produced toward GMP-level specifications. The ambition sounded responsible. In practice, production became costly and slow while the processes were still being developed and could not yet reach that level of stability. Research samples took too long to produce.

The problem, as I saw it, was not a lack of commitment to quality. We were applying the highest specification too broadly and too early. That choice consumed resources without helping every program answer its immediate research question.

We discussed a different allocation. Higher-specification work was concentrated on drugs moving into clinical trials. Other research samples used pilot-production specifications so testing could proceed faster. The change brought production costs into a more appropriate proportion, accelerated research progress, and allowed pilot data to inform process optimization. This was more than a budget correction. Faster access to research samples allowed testing to proceed, while pilot-production data continued to support process optimization at a stage when the process itself was still being developed.

The revised allocation had a clearer decision logic. Clinical materials carried the higher requirement because of where those programs were going. Research samples were produced to answer research questions and improve the process. The sequence mattered because each commitment now belonged to a more appropriate stage.

This is not a general prescription for manufacturing standards; the right specification still depends on intended use, development stage, applicable requirements, process risk, and expert review. The narrower lesson is to avoid funding the final ambition everywhere when the immediate need is interpretable learning for the next decision.

That is why necessary execution and decision-changing work should be named separately. Both can deserve capital, even though one supports delivery and the other tests a choice. Naming the difference keeps the plan from appearing more certain than it is.

One Result Can Change One Decision and Leave Another Untouched

The same result can matter differently to a scientific team, a development committee, a board, an investor, or a potential partner. Each is deciding whether to make a different commitment. Evidence gains strategic meaning through that context.

The antibody program made this visible. Phase I marked clinical progress, but it did not settle the company’s portfolio priority or confidence in future market response. Those questions required evidence and judgment beyond the completed clinical stage.

A single development event may therefore need more than one linked branch map. A technical group may use a result to choose the next study design while a portfolio group waits for additional evidence before comparing the program with other assets. Treating the first decision as resolution of the whole program hides the questions that remain.

A local success may be enough for one group to act and still leave another group without a basis for commitment. Scientific interpretation, financing scope, and commercial judgment do not become contradictory simply because they reach their boundaries at different times. The difference lies in the decision being made and who owns it.

Waiting for every uncertainty to disappear is not a workable development strategy. The better approach is to define a bounded decision now, preserve options where possible, and state what evidence will be required at the next boundary.

A credible milestone statement should therefore identify whose decision it informs, which uncertainty it is intended to reduce, and which material questions remain outside its scope. “Phase I completion will de-risk the program” is too broad. It gives a local result responsibility for manufacturing, commercial, financing, and portfolio questions it cannot resolve on its own.

Preserve the Decision Point in the Capital Plan

A capital plan can describe every activity, dependency, date, and cost while removing the very choice the proposed milestone is meant to inform. If downstream work begins automatically because it has already been approved, the evidence may arrive without a live allocation decision left to change. The branch must remain available.

Begin with the consequential decision and the uncertainty preventing it from being made with acceptable confidence. Then define the smallest credible body of work capable of separating the relevant paths. Smallest does not mean cheapest or fastest. Work that cannot produce interpretable evidence is a poor shortcut even when it delivers on schedule.

Timing belongs in this design. Evidence may matter in principle and still arrive after another commitment has narrowed the available choices. Record when the decision must be made, which dependencies come first, and whether an interim action can be reversed. A reversible next step and an irreversible scale-up should not share an evidence threshold merely because both sit under the same milestone name.

Keep the technical threshold separate from the action criterion. The threshold describes what the result supports within its scientific limits. The action criterion describes what the organization is willing to do with that interpretation, given its alternatives, obligations, and tolerance for waiting. Mixing the two can disguise a managerial preference as a scientific conclusion, or make a technical result look as if it determines a portfolio choice by itself.

When a funding request bundles enabling work, decision-relevant experiments, and preparation for a later stage, separate them. Show which work makes the evidence interpretable, which work is expected to generate that evidence, and which commitments should proceed only after review. A single total may be administratively tidy while concealing the actual sequence of choices.

Make the release point explicit. The organization may decide that continuity, contractual obligations, supply needs, or timing justify funding the full sequence. That can be a legitimate commitment. It should not be described as a preserved decision checkpoint if no later result can alter the allocation.

If no proposed activity can yet resolve the key uncertainty, say so. An enabling tranche can support method development or feasibility work. The honest claim is that it improves the later test, not that the value inflection has already occurred.

End the Review With a Branch Map

Before approving the next tranche of work, require a concise evidence-to-action map. It should name the decision, the owner, the uncertainty being examined, the evidence sought, the plausible result branches, the interpretation boundary, and the corresponding actions. It should also say when the choice must be made.

Keep the map short enough to challenge. Its purpose is not to reproduce the development plan or capture every technical detail. It should show the few assumptions that connect new evidence to a commitment of capital. The relevant scientific, clinical, manufacturing, regulatory, and commercial experts can then test those assumptions from their own domains.

Mark what has deliberately been left outside the current decision. A deferred commercial question or manufacturing dependency does not disappear because it is absent from the present branch. Give it a future checkpoint so a narrow decision does not become an accidental endorsement of the entire program.

If every result branch reaches the same action, record the work as necessary execution or an activity checkpoint. The work may still be required and worth funding; the more limited description simply avoids claiming that it will clarify a choice it cannot change.

Where different evidence would support different actions, confirm two things before funding: the proposed activity can actually distinguish those paths, and the decision owner accepts the branches as real possibilities. Agreement negotiated only after the result arrives does not preserve the original counterfactual test.

A useful sentence to complete is: “This capital will produce [evidence] to clarify [uncertainty], after which [decision owner] will choose among [actions] using [boundary or criterion].” Its value is diagnostic: if the sentence cannot be completed with a clear evidence boundary, the capital request probably still depends on an unstated assumption.

Write the evidence-to-action map before approving the next capital request. If a branch has no credible interpretation, owner, or consequence, leave that decision open. If the work is necessary execution, fund it and manage it on those terms.

At that review, the practical questions are whether the proposed work can genuinely separate the available paths and whether the organization is prepared to act differently when a different result arrives. Reaching the next named stage still matters, but it does not answer those questions on its own.

The aim is not to make every milestone carry a portfolio decision. It is to describe the work honestly. Some work builds capability, some reduces uncertainty, and some changes what the organization is prepared to do next. Calling all three a value inflection hides that difference.

Keep the choice open until the evidence can change it.