Making Stage-Gate and Go/No-Go Decisions

7.1 Designing Effective Stage Gates and Evidence Requirements

A stage gate should occur when management has enough new information to make a consequential decision about further investment. The purpose is not to divide a development process into evenly spaced phases or create recurring executive presentations. Each gate should answer a distinct investment question and determine whether the evidence justifies committing the resources required for the next stage.

Design should begin with the major uncertainties that a program is expected to resolve over its lifecycle. Early work may test whether an underlying scientific or technical concept is plausible. Subsequent stages may establish integrated performance, customer value, regulatory feasibility, manufacturability, scale-up, economics, and launch readiness. The sequence will differ by industry, but the governing logic should remain consistent: investment increases as uncertainty decreases.

For example, an early technology program might pass through gates addressing concept selection, proof of principle, prototype performance, system validation, and commercialization readiness. A therapeutic asset could progress through preclinical evidence, clinical development decisions, pivotal-study commitments, and regulatory or launch gates. A software initiative may use architecture validation, minimum viable product, customer validation, scale readiness, and broad-release decisions. The labels matter less than whether each gate represents a meaningful change in what management knows.

Too many gates create bureaucracy and encourage teams to optimize for reviews rather than learning. Too few gates allow large amounts of capital to accumulate before management formally reassesses the investment case. Gate frequency should reflect the size of incremental commitments, speed of learning, reversibility of decisions, and magnitude of uncertainty. A low-cost exploratory experiment may not require executive approval, while a decision that commits a new plant, pivotal trial, or multiyear engineering organization almost certainly does.

Every gate should have a clearly defined decision question. “Review prototype progress” is weak because it does not specify what management must decide. A stronger question is: “Has the prototype demonstrated sufficient performance and customer value to justify funding integrated system development?” This formulation links the evidence produced by the completed stage to the resource commitment being requested for the next one.

Evidence requirements should be agreed before results are known whenever practical. Otherwise teams can redefine success after disappointing outcomes. Requirements should specify the minimum information expected across the dimensions material to the decision. Depending on the gate, this may include technical performance, reproducibility, customer validation, economic assumptions, regulatory feedback, safety evidence, manufacturing readiness, intellectual-property position, resource availability, or partner commitments.

Good evidence requirements are specific enough to constrain interpretation but not so rigid that management ignores important new information. A requirement might specify demonstration of a defined performance threshold under representative conditions, validation of customer willingness to adopt through agreed research, or evidence that manufacturing yield can reach an economically viable range. The gate should also identify which assumptions remain uncertain even if the minimum threshold is met.

Gate requirements should become more demanding as programs mature. Early programs may appropriately rely on limited experiments, directional customer evidence, and broad value ranges. Later programs should face higher standards for reproducibility, economics, regulatory readiness, operational feasibility, and resource planning because the next investment is larger and the cost of failure rises. Applying late-stage certainty requirements to exploratory work suppresses innovation; applying exploratory standards to late-stage commitments exposes the company to avoidable risk.

Not every criterion should be a pass/fail threshold. Some factors represent preferences that should be weighed alongside the overall investment case. Others are genuinely nonnegotiable. Safety requirements, legal compliance, critical technical performance, or regulatory feasibility may function as hard thresholds. Strategic attractiveness, time to market, or commercial upside may be assessed comparatively. The gate design should make that distinction explicit so that a high score on one dimension cannot inadvertently compensate for failure on an essential condition.

Finally, gates should be linked to funding and resource release. Passing a gate should authorize a defined next increment of work rather than provide unlimited permission to continue. The decision record should state the budget, staffing, facilities, major external commitments, and milestones approved through the next review. This keeps stage-gate governance connected to portfolio prioritization instead of turning it into a parallel process that reviews programs without changing resource allocation.

7.2 Defining Go, Hold, Re-Scope, Partner, and Stop Criteria

A strong stage-gate system offers more than two choices. Binary go/no-go frameworks often encourage continuation because a complete stop can appear unnecessarily severe when evidence is mixed. Expanding the decision set allows leadership to respond more precisely to what the evidence implies. The core alternatives are go, hold, re-scope, partner, and stop, with acceleration sometimes used as a distinct form of go when evidence has improved materially.

Go: A go decision means the program has generated enough evidence to justify the next investment increment. It does not certify eventual success. Management is concluding that, relative to available alternatives, the expected value of continuing exceeds the cost and risk of the next stage. The program should meet required thresholds, retain strategic relevance, have a credible path to value, and possess or be able to obtain the resources required for execution.

A go decision should specify what has actually been approved. If management authorizes twelve months of development through a system-validation milestone, the team should not interpret that as approval of the full commercialization plan. Clear limits preserve management’s ability to reconsider the program as evidence develops.

Hold: A hold is appropriate when immediate continuation is not justified but termination would destroy potentially valuable optionality. The program may be waiting for external data, a regulatory decision, completion of an enabling platform, availability of a critical resource, a partner action, or resolution of market uncertainty. A hold should always specify the condition that will cause the program to return for decision.

Indefinite holds should be avoided. Programs labeled “on hold” frequently continue consuming small amounts of engineering or scientific capacity and eventually become invisible portfolio overhead. The decision record should establish what resources, if any, remain authorized during the hold, who owns the program, and the date or event that triggers reconsideration. If the triggering event does not occur within a defined period, the default action may be termination.

Re-scope: Re-scoping is appropriate when the original program no longer justifies investment but a narrower or materially different version may. This might mean changing the target customer, indication, performance requirement, feature set, technical architecture, geography, or development pathway. Re-scoping should alter the investment case, not simply lower expectations to avoid stopping the program.

A credible re-scope decision should identify which failed or weakened assumption is being addressed and why the revised concept is more attractive. Management should be cautious when teams repeatedly change success criteria after negative evidence. Iteration is inherent in R&D, but continuous redefinition can become a mechanism for preserving programs that no longer meet the original strategic or economic rationale.

Partner: Partnering is appropriate when the opportunity remains attractive but another organization is better positioned to fund, develop, validate, manufacture, commercialize, or scale it. Partnership can include licensing, joint development, co-funding, contract development, strategic alliances, or transfer to another business. It is particularly relevant when a program lies outside the company’s strategic focus yet has external value, or when internal resource constraints would otherwise force termination.

The partner option should not be treated as a way to avoid making a stop decision. Management should assess whether credible counterparties exist, what economics the company could retain, what capabilities remain necessary internally, and how long the partnering process should be allowed to continue. If no viable transaction emerges within the agreed period, the program should return for another decision rather than remaining indefinitely active.

Joint ventures require additional care because the company may not possess unilateral authority to re-scope, partner, or stop a program. Agreements may specify funding obligations, milestone approvals, partner consent rights, technology-use restrictions, intellectual-property ownership, or escalation procedures. Gate criteria for jointly governed programs should therefore distinguish the company’s preferred investment decision from the action it is contractually and practically able to execute.

Stop: A stop decision means that additional investment is no longer justified relative to alternative uses of resources. Common triggers include failure of a critical technical hypothesis, deterioration of commercial potential, loss of strategic relevance, unacceptable regulatory or safety findings, inability to execute, emergence of a superior internal alternative, or an unfavorable combination of remaining investment and expected value.

Stop criteria should be established before emotional and organizational commitment becomes large. Teams should know which evidence would invalidate the investment thesis. These “kill criteria” need not force automatic termination in every circumstance, because unexpected information may legitimately change the context, but deviation from a predetermined criterion should require explicit leadership justification.

The central discipline across all five outcomes is that each decision changes something. Go releases a defined investment. Hold limits activity pending a condition. Re-scope changes the program’s objective or pathway. Partner initiates a specific externalization process. Stop releases resources and closes the program. If the same people, budget, and work continue regardless of the label selected, the stage gate is not functioning as a real decision mechanism.

7.3 Preventing Escalation of Commitment and “Zombie” Programs

R&D organizations are particularly susceptible to escalation of commitment: the tendency to continue investing because substantial resources, reputation, and effort have already been committed. The phenomenon is understandable. Programs create teams, sponsors, technical beliefs, career investments, supplier relationships, and strategic narratives. As these commitments accumulate, stopping a project can feel like admitting that prior decisions were wrong even when continuation no longer makes economic sense.

The first defense is to separate sunk cost from forward value. Money already spent cannot normally be recovered and should not determine whether the next dollar should be invested. Management should ask: if the program did not already exist, would we commit the required resources today based on the evidence now available? If the answer is no, prior expenditure is not a sufficient reason to continue.

This principle does not mean history is irrelevant. Prior investment may have created intellectual property, learning, contractual obligations, equipment, or capabilities that affect forward economics. Termination costs may also be real. Those consequences should be included because they alter future cash flows or options. What should be excluded is the psychological argument that stopping would “waste” what has already been spent.

Zombie programs are a common manifestation of weak stop discipline. These initiatives remain technically active but lack enough resources, sponsorship, or evidence to progress meaningfully. They often survive through small annual budgets because no individual funding request is large enough to attract scrutiny. Collectively, however, they consume scarce specialists, management attention, laboratories, and organizational complexity.

Several warning signs indicate a zombie program:

• Repeated milestone extensions: target dates move several times without resolving the uncertainty the milestone was intended to address.
• Minimum survival funding: the program receives enough money to remain open but not enough to reach a decisive next milestone.
• No clear next decision: teams continue activity without an identified gate, evidence requirement, or investment question.
• Strategic rationale drift: the original rationale no longer applies, and new justifications appear whenever the program is challenged.
• Persistent resource borrowing: the program uses fractional contributions from several functions despite lacking formal priority.
• Unowned accountability: no senior leader actively sponsors termination or acceleration, leaving continuation as the path of least resistance.

Governance design can reduce these behaviors. First, continuation should require affirmative approval. If a gate meeting does not occur or required evidence is unavailable, funding should not simply roll forward by default. Second, programs that repeatedly miss conditions should face increasing scrutiny. A first delay may be normal development uncertainty; a third delay without meaningful learning should trigger reconsideration of the investment thesis.

Independent challenge is also valuable. Program teams should present evidence, but they should not be the sole judges of whether it satisfies the gate. Cross-functional reviewers can test technical assumptions, commercial logic, regulatory implications, resource feasibility, and financial consequences. For especially large or contentious investments, external technical or market experts may help leadership distinguish legitimate uncertainty from organizational optimism.

Decision makers should also examine incentives. Leaders may resist termination if performance evaluations reward project launches, budget size, team growth, or milestone completion without recognizing high-quality stop decisions. R&D organizations should treat disciplined termination as evidence that the innovation system is working. A program stopped promptly after decisive negative evidence may represent excellent management because resources have been released before larger losses occur.

Program closure should be managed professionally. A stop decision does not imply that the team performed poorly. The organization should preserve useful knowledge, intellectual property, data, prototypes, supplier learning, and technical documentation. Employees should be reassigned rapidly to higher-priority work where possible. If termination is treated as organizational failure, teams will rationally avoid generating or surfacing evidence that could lead to it.

Leadership should periodically review the portfolio specifically for zombie programs rather than relying solely on formal gates. Ask which initiatives have consumed resources for extended periods without major evidence, which lack credible sponsorship, and which would not be started today. The objective is not indiscriminate project elimination. It is to ensure that every active program either has an attractive forward investment case or a clearly justified reason for preserving its option value.

7.4 Running High-Quality Stage-Gate Decision Meetings

A stage-gate meeting should be designed as a decision forum, not a project update. The distinction affects who attends, what materials are prepared, how the agenda is structured, and what must occur before the meeting ends. If most of the session is spent explaining project history or reviewing detailed work completed since the last gate, insufficient time remains for evaluating the investment decision.

Participants should include the people with authority and expertise necessary to make the decision. This normally includes the accountable business or R&D sponsor, relevant technical leadership, finance, and representatives from commercial, regulatory, manufacturing, operations, or other functions when material to the gate. The group should be small enough for genuine discussion. Large attendance lists can dilute accountability and encourage presentations rather than debate.

Decision rights should be explicit. One person or formally constituted body should own the decision. Consensus can be useful, but requiring universal agreement gives every function an implicit veto and makes hold-by-default outcomes more likely. Where different leaders control funding, technical standards, or scarce resources, governance should define how disagreement will be resolved.

Materials should be distributed in advance and should focus on changes in evidence since the previous gate. A useful decision package includes the original or latest approved investment thesis, conditions established at the last review, evidence generated against those conditions, updated strategic and economic assessments, principal risks, forward resource request, and program team recommendation. Detailed supporting evidence can be available in appendices or underlying technical documentation.

The meeting should begin with the decision required. The chair might state: “Today we must decide whether to commit the resources required for integrated validation, re-scope the program, hold pending additional evidence, seek a development partner, or stop.” This immediately frames discussion around alternatives rather than inviting an open-ended status review.

The program team should present evidence against previously defined requirements before introducing new arguments for continuation. Reviewers should examine whether the expected learning actually occurred and whether negative findings have been incorporated into the revised case. Material changes to assumptions should be visible. If expected market size has fallen, the technical schedule has slipped, or required investment has increased, those changes should not be buried inside revised forecasts.

Discussion should concentrate on the few uncertainties that could change the decision. Senior teams often waste time debating details on which everyone agrees while avoiding the central trade-off. The chair should ask explicitly: What must we believe for this program to merit the next investment? Which of those beliefs is least supported by evidence? What would cause us to choose a different outcome?

Where disagreement persists, reviewers should distinguish facts from judgments. Participants may agree that a prototype achieved a specified result but disagree about whether the result is commercially sufficient. They may agree on market research but assign different strategic importance to the opportunity. Naming the source of disagreement prevents additional analysis from being commissioned when the real issue is a management judgment that executives must make.

Resource implications should be discussed before the decision is finalized. A go decision is not credible if the required engineers, clinical teams, laboratory slots, capital, or partner capacity have not been secured. The meeting should confirm both the funding amount and the source of critical capacity. If resources remain unresolved, management may issue a conditional decision rather than pretending that the program has been fully authorized.

Joint-venture gates should distinguish internal alignment from formal partner approval. The company should enter the joint decision forum knowing its preferred outcome, acceptable alternatives, negotiation position, and contractual rights. Where the partners disagree, the meeting record should identify escalation mechanisms and any interim limits on spending or development work while the dispute is resolved.

The gate meeting should end with a decision stated aloud and confirmed by the accountable authority. Ambiguous phrases such as “keep moving while we investigate” should be avoided unless they are translated into a defined hold or conditional advance. The approved action, resources, conditions, next milestone, evidence requirements, and review date should be captured before participants leave.

7.5 Checklist: Documenting Decisions, Conditions, and Required Next Evidence

The decision record is the institutional memory of stage-gate governance. Without it, teams reconstruct prior discussions from presentations, emails, or recollections, and conditions agreed at one meeting can disappear by the next. A concise standardized record allows management to determine what was approved, why it was approved, what evidence was expected, and whether the program subsequently delivered against those expectations.

The record should be completed immediately following the gate and distributed to the program team, decision owners, finance, relevant resource owners, and governance participants. It should not attempt to reproduce the entire meeting. Its purpose is to capture the decision and the obligations created by it.

Stage-Gate Decision Checklist

• Program and gate: record the program identifier, program name, current development stage, gate reviewed, meeting date, and accountable sponsor.
• Decision required: state the investment question the gate was intended to resolve.
• Prior conditions: list the evidence requirements, milestones, or conditions established at the previous decision.
• Evidence achieved: summarize what was demonstrated and distinguish completed evidence from assumptions or work still in progress.
• Material deviations: record significant changes in technical performance, schedule, cost, commercial assumptions, regulatory outlook, resources, or strategic relevance.
• Decision: specify go, hold, re-scope, partner, stop, accelerate, or another explicitly defined outcome.
• Decision rationale: state the principal reasons supporting the outcome, including the strongest material concern or counterargument.
• Approved investment: record the budget authorized through the next decision point and identify major capital or external commitments.
• Approved resources: identify critical talent, laboratory, facility, clinical, engineering, manufacturing, or partner capacity committed to the next stage.
• Conditions: specify any requirements that must be satisfied for the decision to remain valid.
• Next evidence: define the technical, commercial, regulatory, operational, or financial evidence expected before the next gate.
• Next milestone: state the milestone, target date, measurable success criteria, and accountable owner.
• Next decision: identify the expected gate date and the investment question leadership will address.
• Dependencies: record prerequisite programs, platforms, partner actions, regulatory events, facilities, or external decisions that could affect progress.
• Override or exception: document any deviation from established thresholds or model recommendations and the executive rationale for the exception.

Evidence requirements deserve particular precision. “Demonstrate technical feasibility” is rarely adequate. The record should state what observable result would constitute sufficient evidence and under what conditions. Similarly, “confirm market interest” should be replaced with an agreed customer-validation standard appropriate to the maturity and economics of the program. Clear requirements make the next gate more objective and reduce opportunities to reinterpret weak outcomes.

Conditions should also have consequences. If a program is authorized subject to recruiting a critical specialist, establishing a regulatory pathway, securing a partner commitment, or achieving a cost target, the decision record should state what happens if the condition is not met. The outcome may be automatic hold, escalation, revised scope, or another formal review. Conditions without consequences become observations rather than governance mechanisms.

For holding decisions, documentation should specify which activities may continue and which must cease. Teams may be permitted to maintain intellectual property, preserve samples, monitor external developments, or complete a narrowly defined experiment while broader development spending is suspended. Funding limits and ownership should be explicit so that the program does not gradually return to active status without a decision.

For re-scope decisions, the revised investment thesis should be recorded rather than simply editing the project plan. Management should understand which original assumption failed, what has changed, what resources the new scope requires, and what evidence will demonstrate that the revised path is attractive. If re-scoping becomes recurrent, the decision history will make the pattern visible.

For partner decisions, record the form of externalization being pursued, the responsible executive, target counterparties or partner characteristics, the resources permitted during the search, and the date by which management will reconsider the program if no transaction is achieved. This prevents “seek a partner” from becoming an indefinite holding category.

For stop decisions, the record should address closure actions. These may include employee reassignment, contract termination, intellectual-property preservation, data archiving, equipment disposition, partner notification, regulatory obligations, customer communication, and capture of technical learning. The portfolio database should also be updated so that released funding and capacity become available for reallocation rather than remaining embedded in obsolete plans.

Finally, the organization should periodically audit whether gate decisions produced the actions documented. If programs marked for termination remain staffed, conditions are repeatedly waived without escalation, or approved priority programs never receive promised resources, the problem is no longer the quality of the gate criteria. It is governance execution. Effective stage-gate management requires a visible chain from evidence to decision, from decision to resource commitment, and from resource commitment to the next evidence-producing milestone.