PSD.13 - Prepare and Return a Decision-Support Recommendation

Type: Method pattern (DPF) Status: Candidate

Primary working result: a decision-support recommendation account for a named recipient, subject, horizon and receiving decision: the supported disposition, considered alternatives, reasons, source limits, material dissent, protected conditions, next use and reconsideration basis. The account may be a short return or a larger package; it informs the later decision without authorizing it.

PSD.13:1 - Problem frame

Use this when analysis is available, or its decisive gap is known, but the recipient still cannot tell what the decision-support performer recommends and on what conditions. A score, workshop agreement or specialist answer is not yet a recommendation for this decision. The same need arises in an ordinary, non-contested request: “Which development direction should we consider over the next four months?”

Start with the receiving question and the strongest answer the available basis permits. Return a supported direction, retained or ranked set, bounded probe, smallest-result request, blocker or abstention. Do not force a single preferred action when the result actually supports a partial order or a missing-premise return.

Problem structuring and decision support is the wider practice. This pattern governs its recommendation-composition and return branch: turning qualified premises into advice usable by a particular recipient. It does not govern the recipient’s choice, the specialist Methods that produced the premises, implementation, or proof that the advice caused an improvement.

The gain is practical: the recipient can distinguish what is recommended now, why, what remains unresolved, what must not be traded away, and what evidence would change the answer. A complete result can be two sentences when they identify one missing premise and the exact recommendation it prevents.

Do not use this pattern merely to reproduce an adequate recommendation, report one specialist result, or make a choice from an already adequate basis. Use the direct communication, supplying-practice or choice guidance instead. An unformed opportunity question needs opportunity construction before advice; an unsettled engagement question returns to PSD.1. Neither case requires a disagreement workshop unless formulation, participation or dissent actually changes the result.

PSD.13:2 - Problem

A recommendation often becomes stronger while it travels. “This option performs better in this model” becomes “the best option”; “the group did not object” becomes “stakeholders agreed”; “the report was received” becomes “the decision was approved.” The lost conditions are precisely those the receiving decision needs.

Development advice adds a second failure. A plausible story can combine human learning, organization change and AI evaluation as if they supplied interchangeable evidence. A course catalogue then stands in for transfer into later Work, a position description for a whole-arrangement comparison, or a benchmark score for deployment safety.

The practitioner needs a positive composition rule, not just warnings. Which conclusion follows from these premises for this recipient? Which alternative remains live? Which missing result is worth requesting? When does declining to recommend supply the most useful answer?

PSD.13:3 - Forces

Force	Tension
Usable direction and warranted strength	The recipient needs an actionable answer; incomplete preferences or evidence may justify only a retained set or bounded question.
Synthesis and supplier ownership	Several results must become one account without absorbing their Methods, evidence limits or authority.
Clarity and material plurality	A short return is easier to use, but omitted dissent or an omitted alternative can reverse its meaning.
Timeliness and protected conditions	Delay has consequences, but urgency does not convert missing permission, consent or safety evidence into an admissible premise.
Professional judgement and recipient choice	An adviser should explain a reasoned position without making the recipient’s values, mandate or commitment for them.
Continuity and changed conditions	An earlier recommendation may remain useful in part even when one configuration, source or assumption changes.

PSD.13:4 - Solution

Compose the recommendation at the strength and scope of its decision-bearing premises, then return it for the agreed use. The following moves express information dependencies, not a compulsory sequence of meetings or documents. Adequate existing results may close several needs at once.

PSD.13:4.1 - Recover the receiving use and the right to provide this advice

State who receives the answer, the exact subject or holder, the horizon, the question and the later decision or other Work that will use it. Recover the actual choice owner separately. A recipient may also be the choice owner when that authority is established; receiving an answer does not establish it.

Use an adequate PSD.1 engagement result for the same subject and decision. Otherwise obtain the qualified direct facts or return the exact missing recipient, question, scope or authority premise. Do not infer the engagement from a title, course, department or tool output.

For development-direction advising, distinguish the recommending performer from the recipient at the grain of this application. The client, sponsor, holder, recipient and chooser may coincide in some respects and differ in others; recover only the differences that affect the advice, disclosure or later choice. A larger organization may contain both adviser and recipient. Merely naming two roles of one self-advisory actor does not create the distinct-performer engagement; direct domain and C.11 guidance remain available.

Establish applicable professional scope, competence, material conflicts and permitted reliance before offering positive advice. For example, management-consultancy engagement and conflict conditions are informed by the ICMCI competence framework v4.0, C.1.1–2, C.2.1 and E.3.5. Human career-service consent and assessment limits come from the applicable professional source, such as the NCDA 2024 Code of Ethics, A.2 and E.2. These are different service conditions, not one universal profession or evidence that a proposed intervention works.

If a material mandate, competence, conflict or admissibility condition cannot be satisfied, return the scoped blocker or appropriate referral. A persuasive presentation cannot repair that condition. Existing emergency, safety or legal response duties retain their own direct authority and must not wait for completion of a PSD account.

PSD.13:4.2 - Recover what the available premises actually support

For every premise that can change the disposition, recover its claim, exact subject and configuration, source or supplier, applicable window, intended use, uncertainty and reliance limit. Inspect available results first; use A.15.9 to reuse an adequate result, retain its useful limit, or select a worthwhile obtainable missing contribution, and A.10 to qualify the actual reliance. This pattern adds no second specialist-acquisition Method.

Read a comparison as a result over particular alternatives under particular values, evidence and assumptions. A PSD.12 robustness result contributes only where it can change this recommendation for the same configuration and horizon. It neither entails the whole recommendation nor authorizes the later decision. Preserve the PSD.9–PSD.11 comparison basis needed to interpret it.

Keep the live set and the compared subset explicit. A result about internal development and one mixed human–AI arrangement does not dispose of an external provider or a materially different probe. A gap about one candidate can prevent ranking that candidate while leaving an independent comparison usable. Do not narrow the set simply because evidence is easier to obtain for the familiar option.

Preserve material agreement, disagreement and affected-party concerns from PSD.7 or a qualified direct account. Distinguish a factual objection, a value conflict, an authority condition and a question about representation; they call for different responses. Give the objection and its consequence enough prominence to survive the return. Attribution and disclosure must respect the applicable protection and confidentiality conditions.

For development opportunities, consume a qualified opportunity result: the problem or promise, dependencies, bounded reachability, uncertainty and next question. A search/archive entry, a fashionable technology or a candidate label supplies none of those by itself. Opportunity construction is a separate Method with an independently useful result; composing advice does not complete missing search, dependency or reachability work.

PSD.13:4.3 - Select the warranted recommendation disposition

Ask what the recipient can responsibly consider now, given the question, alternatives, protected conditions, comparison and consequence of delay. Explain the inference from the relied-on premises to the disposition. The six forms below are result branches, not stages to complete.

Disposition	Positive basis and useful return	Boundary
Retained or ranked set	Several directions remain live. Return the warranted order, partial order or trade-offs and the unresolved distinction that could narrow it.	A retained set is not consensus; an order valid under one scheme is not a universal ranking.
Supported direction	Qualified premises and a declared decision-support rule justify preferring one direction for this use. State the advantage and the conditions under which it holds.	Other candidates need an explicit disposition; protected conditions cannot disappear into a compensating score.
Bounded information-gaining probe	A specified uncertainty could change the advice, and a qualified probe is feasible and proportionate to the delay, burden and risk it introduces. Name the question, bounded exposure, useful evidence and stop.	Thin evidence alone does not justify a probe. The probe’s authorization, WorkPlan and execution remain separate.
Smallest-result request or handoff	A named practice can supply a missing comparison, feasibility, interpretation, or other result whose obtainable contribution warrants the full formulation, qualification, delivery, performance, interpretation, delay, and displaced work for this receiving question. Use A.15.9 and C.11.DUA for that selection.	Request the bounded premise; the supplier keeps its Method and does not take over the recommendation or choice. A missing premise alone can remain a limitation.
Blocker	A missing, stale, incompatible, inadmissible or unauthorized premise prevents the requested responsible recommendation. Name the premise, affected use and what would resolve or reroute it.	State any still-usable partial result; do not convert uncertainty into either a confident negative or a positive recommendation.
Abstention	The recommendation premise itself fails, or the requested kind of advice cannot responsibly be supplied within the engagement. State the failed premise and the correct receiving account where one exists.	An abstention is not an adverse factual finding about the subject and is not abandonment of any obtaining professional duty.

A mixed answer can be useful if its scopes are unmistakable: retain two directions and block a stronger ranking, with a comparison request only when selected for its attainable value. Do not present these as contradictory overall statuses.

The strength of each claim stays inside the narrowest material limit of the premises on which that claim depends. If one input supports only a particular AI configuration, the recommendation cannot silently extend it to another. If an unused source expires, it does not invalidate unrelated claims merely by appearing in the same bibliography.

When a total order is unwarranted, prefer an explicit partial answer to invented weights or probabilities. When the declared scheme does support one direction, do not conceal that conclusion behind an uninformative list. Explain why a serious remaining alternative loses for this use, and what would reverse that judgement.

PSD.13:4.4 - Make the recommendation account inspectable at the smallest useful size

The account is a claim-bearing result, not a mandatory template or new record kind. A short message may carry it; a consequential institutional matter may need a larger package and protected annex. Keep the following content recoverable wherever it can change reliance.

Position	Decision-useful content
Receiving frame	Recipient, recommending performer, holder or subject, question, horizon, choice owner, receiving decision and next Work.
Candidate domain	Considered alternatives, compared subset, exclusions with their basis, retained alternatives and any candidate gap.
Relied-on premises	Exact results and relevant source editions or dates, configurations, qualification windows, evidence limits and supplying owners.
Comparison and inference	Scheme, material trade-offs, protected conditions, robust regions or reversals, and the reasoning that supports the disposition.
Unresolved content	Material uncertainty, dissent, evidence losses, unsupported extensions and what each prevents.
Return	One or more explicitly scoped dispositions and the next consideration, request, reroute or separately governed choice they support.
Reconsideration	The source, observation, configuration, threshold, authority, horizon or candidate change that could alter a specified claim or disposition.

Put the disposition and its most important limit first. Keep the evidence and reasoning reachable without requiring the recipient to reconstruct the entire analysis. Where a protected annex limits access, the accessible return must still state the resulting reliance restriction; it must not hide that a decisive premise is unavailable to the recipient.

A threshold in this account needs a source and an owner for its meaning. “Reconsider when recovery evidence no longer supports the continuity condition” is useful only if the named evidence, condition and receiving question can be recovered. PSD.14 designs how such evidence will be obtained and interpreted. If a positive recommendation depends on ongoing observation, an unassigned or infeasible observation obligation is a material limit now, not an administrative detail to repair after handoff.

PSD.13:4.5 - Return the advice without taking the later choice

Make the result available to the actual recipient in a form suited to the agreed use. For a receiving-use question, apply A.2.9: what should the recipient understand or do, what evidence is enough to judge that, and what smallest repair or stop follows? Explaining the conditional nature of a ranking may be necessary; obtaining agreement with it is not the same task.

Keep the communicative Work, recommendation content, carrier, recipient response and later effect distinct. Delivery establishes neither understanding nor reliance, consent, choice, implementation or effectiveness. A recorded acknowledgement may support receipt; it does not establish an authorized decision. A communication can perform an authorization only under its own actual performer, authority and institutional conditions.

The authority-holding System may later choose, reject, retain alternatives, commission a probe or reroute under C.11 or its direct domain rule. A corporate organ may consume the recommendation and unresolved disagreement through CGOV.11; that organ, not this account, performs the authorization Work. Do not require a corporate-governance framework where the actual decision is personal or otherwise outside its subject.

Returning the advice closes this recommendation application, including an honest stop branch. Later premise refresh and a newly composed recommendation are later work. Any promised professional contact, referral or service closure remains governed by the actual relationship; closing this application neither invents nor cancels those duties.

PSD.13:5 - Archetypal Grounding

The examples are hypothetical. Assumed specialist results illustrate the pattern; they assert no real public decision, client assessment or intervention effectiveness.

PSD.13:5.1 - A pump comparison that cannot yet choose an investment

An East-District authority asks a separate analysis team for advice on flood-pump investment for the coming season. Its receiving decision is whether and how to commit funds. The live candidates include a baseline N, fixed pumping F, mobile pumping M and staged combination S.

The qualified numerical slice concerns pumping-service loss and incremental cost, not the whole public consequence. In the stated model, F has loss 2 in normal access and 3 with road loss, at cost 8; M has loss 1 and 9, at cost 5. The baseline and staged alternative are not closed by this two-option comparison. Reachable assistance, property consequences and protected conditions still require their direct evidence and value judgements.

A sensitivity result supplies the conditional boundary 7p > 1 + 3λ for F to have the lower weighted loss, where p is the assumed road-loss probability and λ the declared cost weight in this illustrative model. Neither value has been established for the actual investment question. With λ = 0.5, p = 0.2 favours M, while p = 0.6 favours F. This is evidence of a possible reversal, not evidence that either probability obtains.

The team returns:

Retain fixed and mobile pumping for the current pumping-service/cost comparison. A whole-investment recommendation is blocked by the unqualified road-access basis, unresolved assistance and protection consequences, and the unclosed baseline/staged branches. The current model favours different options under different declared assumptions; it does not justify a preferred investment. Reconsider the affected comparison when those results or the authority’s value and protection premises change.

A material objection about reachable assistance remains in the return, not buried below the numerical result. The authority can decide what further evidence to commission under its own mandate. A bounded consequence/access request is a useful continuation when an obtainable result can improve the investment decision enough to warrant its acquisition burden. Neither that request nor the conditional comparison authorizes spending, a public trade-off or an operational response.

PSD.13:5.2 - Development advice with a positive branch and an honest gap

A committee asks a separate advisory team which direction to consider for ninety-day service reliability. The holder is the service organization; the committee is recipient and, by the assumed mandate, choice owner. Internal development, an external provider and a mixed human–AI arrangement are live directions. Critical-service continuity, a security boundary and a bounded budget are protected.

An operations account, position/interface account and strategy priority are available, but none compares the whole obtaining arrangements. The first complete return retains the three directions for consideration and states that their whole-arrangement ranking is unsupported for this horizon and protected conditions. A published Method for obtaining a comparison is not the client-specific result. A comparison request follows only if its attainable contribution warrants the full acquisition burden; the unranked answer need not propose that work.

For a separately stated continuation, suppose qualified allocation and security results now cover all three directions. They support a bounded mixed-arrangement probe, retain internal development, and exclude this provider configuration under the security condition. The probe evidence question, feasibility, exposure and stop are supplied rather than invented. The adviser can then recommend considering that probe first while retaining internal development; throughput remains uncertain and the protected conditions remain binding. These new premises, not a change of wording, permit the stronger answer.

This continuation does not promote an earlier I/H-only comparison into full-set closure. A different provider, a materially different probe, or model configuration B needs its own qualification. If the mixed result covered only configuration A, advice about B retains that qualification gap or a blocker while independently qualified internal-development content can still be returned. The committee’s later choice and probe WorkPlan are separate.

A material provider interest must be disclosed and treated under the engagement’s conflict conditions before this positive advice is relied on. A sponsor wanting one supplier does not settle the holder’s need or the committee’s choice.

PSD.13:5.3 - Unlike holders and ordinary non-use

For a person asking a separate adviser about four months of engineering-management development, suppose qualified accounts of current work show useful cross-discipline coordination experience, while an additional individual transfer study would consume much of the available horizon. The adviser can recommend considering a coordination direction within the person’s accessible work and support, with uncertainty about the four-month capability gain. This is a substantive direction for consideration, not proof of intervention effect or independent transfer. A course attendance record alone would not support that advice. When current premises support two directions without ordering them, return that retained set and its distinguishing limit; the person’s later choice remains theirs.

For an authorized team considering an AI scaffold or human–AI allocation, name the exact changed object, version, environment, evaluation validity, oversight and protected conditions. Use AI/security and, where relevant, organization-allocation results. A human learning study or a ranked recommender-system output does not establish those premises. If a known inadmissible configuration cannot meet the protected boundary, block that recommendation; if the evidence is merely absent, name the gap rather than claiming failure.

For a non-cultural population or lineage with no population-local recipient or chooser, abstain from attributing a recipient-owned development recommendation to it. Return population change to its evolutionary account. A separately authorized research team considering an intervention is a different recipient-owned case with scientific, safety and governance premises; naming that case establishes none of them.

For a chooser who already has an adequate option set and needs only to make their own decision, use direct domain or C.11 guidance. Neither an adviser nor a PSD return package needs to be invented.

PSD.13:6 - Bias-Annotation

A decisive-sounding answer can conceal uncertainty, but perpetual qualification can also conceal a warranted recommendation. Test both errors: what stronger claim is unsupported, and what useful claim is being withheld despite adequate premises?

Sponsor preference, provider interest and model convenience can shape the candidate domain before the recommendation is written. Inspect exclusions and unclosed branches, not just the displayed winner. Preserve material dissent without implying that every view has equal evidential support or that unanimity is required.

An available numerical score may dominate attention while a consent, continuity or access condition is left in prose. Keep non-compensable conditions visible at the point of disposition. Conversely, do not invent a protected condition from the adviser’s personal preference.

PSD.13:7 - Conformance Checklist

Recognition asks whether recommendation composition is the missing result: a named recipient needs an answer that combines qualified premises, and the account can change their next consideration. A short supported answer with its necessary limit can close a small use; a selected request or exact blocker can also be useful.

Assurance asks whether this particular disposition is warranted for its intended reliance. Consequential uses add the direct domain’s evidence, competence, independence, consent, safety, security, institutional and validation conditions; the size of the account does not lower them.

Check	Sufficient answer
Receiving use	Recipient, subject, horizon, recommending performer, receiving decision/Work and separate choice boundary are recoverable.
Premise fit	Every decision-bearing input fits its actual configuration, window and use, or its gap and consequence are explicit.
Candidate and comparison coverage	The live set, compared subset, retained alternatives, exclusions, scheme and material reversals are visible.
Warrant	The inference supports the stated disposition without inventing preferences, probabilities, authority or supplier results.
Conditions and dissent	Material protection, disagreement, evidence loss and professional restrictions survive the return.
Use and reconsideration	The recipient can find the next use and exact reason to reconsider; required observation is feasible or named as a limit.
Result boundaries	Recommendation, communication, later choice, WorkPlan, implementation and effect have not been substituted for one another.

If one check fails, repair or narrow the affected claim or return the honest stop. Request a missing result only when its obtainable contribution warrants acquisition. Do not erase independently usable content to make the account look uniform.

PSD.13:8 - Common Anti-Patterns and How to Avoid Them

Anti-pattern	Repair
Copy the highest score into “we recommend”	Recover the candidate domain, preference/evidence conditions and decision-support inference; retain a partial order when that is all they support.
Make a stronger recommendation by omitting an inconvenient branch	State the compared subset and the unclosed alternative or objection, with its effect on the answer.
Treat a specialist answer as the whole advice	Preserve supplier scope and compose only the inference that this recipient’s question still needs.
Use a probe as the default answer to uncertainty	Name the decision-changing question, feasible evidence, exposure, delay and stop before recommending the probe.
Turn delivery or agreement into authorization	Obtain the later choice and authority evidence separately; repair understanding without seeking manufactured consent.
Offer a universal development prescription	Recover holder-specific premises; return the smallest gap where human, organization and AI evidence differ.
Hide a blocker behind a long report	Lead with what cannot yet be responsibly recommended and the exact result that could change it.

PSD.13:9 - Consequences

The recipient gains a bounded answer that can be challenged, declined, acted on through the proper decision, or refreshed without reconstructing the entire engagement. Partial results and honest stops become usable outputs rather than signs that the analyst failed to choose.

The cost is explicit reasoning at the boundary between analysis and advice. Some requests cannot be closed with a single direction, and some attractive probes remain unsupported. This preserves the distinction between useful decision support and a persuasive but unwarranted prescription.

PSD.13:10 - Rationale

Recommendation composition has work left after generic evidence, comparison and choice guidance. Evidence can support a claim without determining what to advise; a comparison can leave alternatives unresolved; a choice rule belongs to the later decision. This pattern connects those qualified results to one recipient’s question and makes the supported disposition intelligible.

The six return forms avoid two extremes: treating every uncertainty as failure, and treating every engagement as a duty to name a winner. They do not replace domain decision rules. The narrowest-material-premise rule applies claim by claim so that one failed branch neither contaminates unrelated advice nor disappears inside an overall positive answer.

PSD.13:11 - SoTA-Echoing

The practice question is how to return useful advice from qualified analysis without manufacturing a complete order, professional competence or recipient choice. The selected line combines explicit decision-analysis warrant with a source-bounded professional return. Its additional explanation costs effort; that cost is justified where a lost condition could change the next decision, and is reduced to a short return in a simple case.

Question and selected move	Serious alternative, defect and changed locus	Source role, limit and reopen condition
How does comparison become advice? Adapt structured preference-to-recommendation reasoning, preserving the actual problem and model conditions.	A single score is economical when a justified model covers the question. It fails when incomplete preferences or an unclosed candidate branch are silently filled in. Sections 4.2–4.4 require the scheme, inference and scope rather than banning scalar comparison.	Greco, Słowiński and Wallenius (2025), Fifty years of multiple criteria decision analysis, is a best-known-line synthesis of finite-action MCDA and different preference-to-recommendation procedures. It supplies neither this engagement’s values nor universal superiority of one procedure. Reopen when a rival gives equally warranted advice with materially lower burden, or the finite-action/model assumptions cease to fit.
What must accompany technically supported organizational advice? Adapt explicit engagement, client involvement and conflict qualification.	A technically correct memo can suffice for a bounded analytical request. It is insufficient when reliance also depends on mandate or a material provider interest. Sections 4.1 and 5.2 keep those conditions with the advice.	ICMCI v4.0 (2021), C.1.1–2, C.2.1 and E.3.5, is a substantive professional comparator, not a universal procedure or effectiveness study. Reopen on a changed applicable service rule or evidence that this qualification omits a material relationship condition.
What makes a human career return usable? Adopt within that service intelligible basis, consent and competent assessment use; retain referral where required.	Generic “best next course” advice is shorter but can conceal the person’s receiving question and unsupported assessment. Sections 4.1, 4.5 and 5.3 retain these boundaries while leaving intervention and transfer evidence to their owners.	NCDA Code of Ethics (2024), A.2, A.10 and E.1–2, is a profession-specific normative comparator. It proves no learning effect and supplies no AI-holder assessment rule. Reopen when the actual service or applicable professional conditions change.

PSD.13:12 - Relations

PSD.1 supplies the bounded engagement question for the same subject and receiving decision. Its availability does not establish the particular mandate or impose temporal order.
PSD.7 supplies material shared and contested claims. They inform the recommendation without settling the later choice.
PSD.8–PSD.11 supply the candidate, value, uncertainty and comparison results needed to interpret the advice. PSD.12 supplies robust regions, reversals and information priorities only where they can change this recommendation for the same configuration and horizon.
A.15.9 supplies qualified use and selection of worthwhile acquisition for external results; A.10 governs their actual reliance. Holder practices retain facts, Methods, protected conditions and conclusion authority.
A.2.9 supplies receiving-use judgement and the distinction between communication and its institutional effects. C.11 or the direct domain rule owns the later choice; WorkPlan, Work and effect remain separate.
PSD.14 uses this recommendation to establish which later decision, consequences, signposts and premise changes matter. It obtains later authorized decision and implementation evidence as distinct results; this recommendation supplies neither.
CGOV.11 may use the recommendation and unresolved disagreement for the named corporate matter. Only the authority-holding System supplies the authorization Work.
Missing, stale, out-of-scope or incompatible contributions call for an adequate qualified direct result or the exact missing-result return. A neighboring pattern, profile or source being available does not supply a performed result or require a complete language traversal.