Different audiences can prefer different portfolios

Different stakeholders can understand the same evidence and legitimately prefer different actions. A strategy sponsor may prioritise capacity outcomes, an operating owner may prioritise executable improvement and a finance function may emphasise the burden of initial capital. The analyst should distinguish this preference disagreement from disagreement about performance. The model can calculate the consequence of each supplied vector. It cannot determine which participant has authority to govern the commitment or whether an apparent compromise is legitimate.

A stakeholder scenario needs a mandate, not merely a role label. It should state the objective, commitment stage, represented interests and decision rights associated with the perspective. A finance-labelled vector does not automatically represent the organisation's financial policy; a delivery-labelled vector does not establish what the delivery team actually believes. Actual participants must supply and authorise their interpretation. The illustrative mandates in this paper are authored alternatives designed to test conditional preference, rather than survey findings or descriptions of an institution.

Authority matters because averaging can conceal a decision that still needs to be made. If a board has delegated implementation priorities to an operating committee within a strategic envelope, a simple average of unrelated stakeholder vectors may override that delegation. If affected groups have a consultation or consent role, an average may dilute rather than satisfy the relevant process. The governance review should identify whose preferences inform the analysis, whose approval is required and who can adopt the final mandate. The calculation follows that structure; it does not create it.

The team should also distinguish advisory diversity from collective choice. Separate stakeholder scenarios can reveal which proposals remain attractive across perspectives and which depend on contested priorities. That is useful diagnostic information even if the committee ultimately adopts a single authorised vector. A collective choice process may then negotiate, deliberate or use a formal rule. Each approach has its own legitimacy conditions. Presenting the arithmetic mean of weights as consensus skips the work of establishing whether participants accept the tradeoffs embedded in that mean.

The illustrative balanced mandate prefers Reliability architecture, the impact-led mandate prefers Capacity extension and the initial-capital mandate prefers Commercial learning. No candidate performance changes across those calculations. The comparison establishes that different supplied tradeoffs can produce different leaders. It does not establish which tradeoff a real committee should adopt, and the initial-capital case still does not solve a constrained funding allocation.

The mandate reversal is an opportunity for a focused conversation. Participants can inspect the raw performance and explain which swing they value differently. If a strategy sponsor prefers the capacity proposal because the decision is intended to secure a future service capability, the committee can examine whether the readiness-heavy baseline fits that purpose. If an operating owner prefers the reliability proposal because the present commitment concerns a near-term service obligation, the team can test whether the alternatives belong in different stages. The conflict may require a clearer charter rather than a compromise weight.

The finance perspective needs particular care. Greater cost preference can reflect a desire to preserve capital, but it cannot enforce an approved envelope or account for expenditure timing by itself. A stakeholder may believe that a proposal is unaffordable, which is a feasibility claim, rather than simply less attractive on cost. The facilitator should place that claim in the appropriate part of the decision process. Mixing an affordability constraint into a cost weight can obscure the reason an option is rejected and leave the supposed constraint open to compensation.

A common performance table helps keep the discussion disciplined. Participants should be able to disagree about priorities while using the same scope and evidence interpretations. If they also dispute raw ratings, the team should record that as a separate channel of disagreement. A combined stakeholder scenario can be legitimate, but it should identify both changes explicitly. Otherwise a preference dispute may be misread as a technical difference, and participants may talk past one another while comparing totals generated from different premises.

The facilitator should look for agreements that do not require identical weights. Participants may agree on eligibility, acknowledge that an option is dominated on represented attributes or support a bounded investigation before implementation. They may also agree on the evidence needed to resolve a performance dispute. These agreements can support progress while priorities remain contested. The aim is not to force every participant into a single vector. It is to identify what action is authorised now and what disagreement still governs a later commitment.

A robust shared preference can be useful when it is actually demonstrated over relevant perspectives. If the same option leads under several authorised scenarios, the committee has evidence that the present disagreement does not alter that particular preference. It should still state the perspectives tested and the remaining constraints. In the frozen mandate experiment, the leaders differ, so the paper cannot claim cross-mandate agreement. The result directs attention to the decision envelope and authority that select the applicable vector.

Compromise can still be appropriate. Participants may deliberately adopt an intermediate vector after understanding the tradeoffs and accepting the resulting action. That is a negotiated judgement, not an objective discovery by averaging. The record should explain why the compromise is acceptable, what interests it represents and what conditions would reopen it. The analyst can compute its consequence once it is supplied. The committee should avoid retrospective claims that the compromise was inevitable or that dissent vanished because a number was agreed.

Dissent should be preserved in a form that can improve the decision later. A dissenting participant can state the alternative mandate, the preferred action under it and the trigger that would make its concern decisive. For example, a capacity-oriented perspective may become governing if the organisation formally changes the commitment stage to long-term service expansion. A delivery-oriented concern may become decisive if a required capability fails. This is more useful than recording only that someone opposed the final choice. It gives future reviewers a proposition to examine.

Power asymmetry can shape elicitation. Participants may hesitate to challenge a senior sponsor's weights, or a facilitator may assume that agreement in the room means informed acceptance. A credible process creates a way to state alternative scenarios and unresolved concerns without treating them as obstruction. It can ask participants to explain their own rationale before seeing a consolidated result. The exact process depends on the institution, but the methodological requirement is clear: the model should not make contested authority look like technical consensus.

A mandate dispute can be diagnosed by asking each participant to interpret the same opposed pair rather than another complete ranking. The impact-led participant explains the advantage for which greater commitment is acceptable; the liquidity-led participant explains which commitment difference it refuses to exchange. If both describe the same decision horizon and resource, the disagreement concerns preference. If one is selecting a development study while the other is protecting an implementation envelope, the apparent weight dispute concerns different actions. Repair that mismatch first. The supplied mandate reversals reveal conditional preferences; they cannot establish that averaging the vectors would create a coherent shared action.

A retained minority scenario is most useful when attached to a prospective switch condition. An impact-led alternative might become relevant when the institution moves from near-term delivery to an authorised expansion decision. That is different from a minority claim that the current impact evidence is understated: the latter requires a corrected performance interpretation, not merely another vector. Preserve the alternatives with these different triggers. At review, test whether the triggering change actually occurred before presenting the old minority ordering as current advice. This keeps dissent actionable and prevents an archived scenario from becoming a convenient substitute for resolving the disputed proposition.