The option universe is chosen before scoring

The option universe determines what a preferred result can mean. A calculation can identify the highest score among the supplied candidates without establishing that an important alternative was considered. If the universe consists only of proposals promoted by teams with budget influence, the result may reward organisational access rather than better ways to achieve the objective. The committee needs a record of how alternatives entered the comparison and why the search stopped. A complete spreadsheet is not evidence of a complete search.

Option development should start from the problem and the required outcome. A request to replace an operating system may conceal a need for more reliable information, faster response or lower support burden. The organisation could address that need through a different system, a change in process, a service arrangement or a narrower intervention. Starting from the preferred solution makes those alternatives appear peripheral. Starting from the outcome gives them a fair opportunity to demonstrate whether they can meet the same need with a different commitment.

The search should distinguish an option from a variant. A supplier configuration, a delivery phase and an operating model can each be described as alternatives, although they answer different questions. Mixing them in a single list can give a large concept several chances to win while representing a competing approach only through an inconvenient variant. The analyst should define the level of comparison. If the decision concerns an operating approach, the shortlist should compare operating approaches. Supplier variants can be evaluated within the selected approach through a later, appropriately specified procurement decision.

A credible universe includes the reference action against which intervention is justified. Continuing current arrangements may carry deteriorating service, maintenance expenditure or exposure to an unmet obligation. Doing the minimum necessary to remain admissible may require an active programme rather than literal inaction. Those distinctions matter when costs and benefits are compared. The reference is not a free option with no consequences. Its scope should be described with the same discipline as the proposed changes, including what would happen without discretionary expansion.

The reference also establishes a causal question for benefits. A proposal should not receive credit for improvements that the organisation expects to obtain through an already committed intervention. Nor should its cost perimeter omit enabling work that would not otherwise occur. The analyst should identify what is incremental to the reference action and what remains common to every alternative. The Portfolio compass does not perform that incremental analysis. It receives supplied raw values. A consistent reference is therefore part of the evidence architecture preceding the calculation, rather than a result that the score can establish.

Inclusion rules should be stated before the team observes a convenient winner. A rule may require a credible route to meeting the service objective, a defined commitment perimeter and enough evidence to permit a meaningful exploratory comparison. It should identify who can nominate an alternative and how an omission can be challenged. The rule need not admit every conceivable concept. It should distinguish exclusion because an option cannot address the problem from exclusion because its sponsor has not yet developed the evidence needed for this decision stage.

A concept with incomplete evidence may belong in a development register without entering the immediate commitment shortlist. That treatment preserves the idea while preventing an untested promise from competing as if it were executable. The record should state what evidence would make the concept eligible for reconsideration. Otherwise exclusion becomes permanent by administrative inertia. A sponsor should be able to distinguish a rejected mechanism from an incomplete dossier. The former calls for a different idea; the latter may justify targeted work if the potential decision consequence is material.

A feasibility screen should not disguise an outcome preference. Rejecting an option because it is unlikely to score well on strategic impact is different from rejecting it because it cannot meet a required service condition. The former anticipates the comparison and can eliminate a meaningful tradeoff. The latter applies an admissibility rule. The distinction is particularly important when a low-cost or staged intervention provides less benefit than a larger project but remains a legitimate response to the objective. The screen should retain it unless the required outcome truly makes it inadequate.

Early dominance can simplify the search, but it needs an evidentially fair comparison. An option that is no better on any relevant attribute and worse on at least one may not need elaborate preference analysis. The conclusion depends on common definitions, comparable evidence and the absence of an omitted distinguishing feature. A less mature proposal should not be declared dominated merely because its benefit has not yet been established. Unknown performance is a reason to investigate, bound or exclude under an explicit evidence policy. It is not automatically the worst possible performance.

The search record should preserve rejected alternatives and their reasons. Keeping them visible allows a later reader to distinguish a considered option from a forgotten one. It also supports a sensible reopening process. A previously infeasible service arrangement may become credible after a supplier develops relevant capacity. An expansion concept may cease to be viable after the organisation changes its strategy. The decision universe is bounded at a date for a purpose, not frozen for the lifetime of the organisation. A material change in that boundary needs a new comparison record.

Changing the universe can change the arithmetic even when the original proposals remain unchanged. The canonical app uses the supplied candidates to set the observed ranges for normalisation. Adding a very costly proposal can compress the normalised cost differences between existing candidates. Adding an option with weak readiness can change the meaning of every unchanged readiness value. A score increase after the universe expands does not establish improved performance. It may reflect a different scale. The worked set-extension study later makes that mechanism explicit through frozen inputs and outputs.

This dependence creates a governance reason to preserve a stable candidate set during a comparison session. A team can revise the set when an inclusion error is found, but the change should trigger a fresh interpretation of scores and weights. Earlier score levels should not be presented as directly comparable with the new levels. The raw evidence remains useful for continuity. The normalised scores belong to the stated set. Keeping both views allows the committee to understand whether a changed recommendation arises from new performance evidence, a changed universe or a revised preference.

The organisation should not insert a poor candidate merely to make preferred proposals look stronger. A dominated reference can alter observed ranges while contributing no credible choice. This is not evidence that range normalisation is always unusable. It is evidence that the model's scale is partly governed by admission decisions. The appropriate repair may be a tightly defined eligible universe, a separate experiment with fixed external anchors or a method that does not use candidate extremes. The app implements the observed-range method; alternative scaling architectures in this paper remain conceptual unless a separate verified model establishes them.

A strong search process invites a challenge to the mechanism, rather than asking reviewers only to propose a different supplier. An operator might ask whether the organisation could change the demand it needs to serve, share a resource with another activity or remove a bottleneck through a smaller intervention. These ideas can alter the strategic option space. The challenge should remain proportionate: a committee considering a near-term operational response cannot indefinitely defer action while pursuing speculative alternatives. The record should state the search effort justified by the scale, reversibility and uncertainty of the commitment.

Closing the universe requires a reason that can be reviewed. A fixed decision date, a material cost of delay or diminishing value from additional search can justify closure. Convenience alone is weak justification when an omitted mechanism could plausibly change the commitment. The analyst should record significant unresolved alternatives and the condition under which they would reopen the decision. This does not make the recommendation indecisive. It gives the committee a controlled way to act now while recognising that the set of considered possibilities is an analytical boundary.