Model Role SOP / Operating Playbook for Management Consulting

Purpose and scope

This playbook helps a management consultant convert an uncertain organizational question into a bounded, evidence-led decision process. It connects intake, problem framing, discovery, qualitative and quantitative analysis, option development, recommendation delivery, implementation handoff, measurement, escalation, and learning.

The playbook applies to advisory work such as operating-model, process, performance, strategy, transformation, technology, risk, organizational, customer, or market questions when those activities fall within the signed engagement. It is designed for a consultant who may own a workstream or deliverable while the client retains organizational decision authority.

The model does not authorize the consultant to:

• expand scope, budget, access, deadlines, deliverables, or contractual commitments;
• make the client’s executive, financial, employment, legal, regulatory, security, privacy, product, operational, or investment decision;
• access or disclose information beyond the approved need and permission;
• implement a change in a client system unless that action is explicitly authorized and controlled;
• represent a modeled benefit, estimate, hypothesis, or generated output as an observed fact; or
• use a method, framework, dataset, AI tool, connector, or external service merely because it is technically available.

Evidence-derived role standard

The vacancy evidence supports a role centered on analysis and diagnosis, workstream leadership, client discovery, facilitation, research, recommendation delivery, and implementation readiness. In the observed sample, analysis and diagnosis appeared in 64 of 100 vacancies, workstream or engagement leadership in 53, client or stakeholder discovery in 27, workshop facilitation in 20, and research as an explicit duty theme in 12. These themes overlap; an unstated duty is unknown rather than absent.

The most visible work products were recommendations and insights, plans and roadmaps, reports and documentation, executive presentations and briefings, models and dashboards, and playbooks or operating procedures. The evidence also supports decision memos, current-state assessments, gap analyses, process maps, business cases, risk registers, stakeholder maps, implementation plans, action trackers, and quality-review records.

The separate current-changes study adds eight operating implications: define the research boundary before using source-aware tools; inspect conversational analysis and retain an audit note; preserve source authority and lineage; decide confidentiality and permission architecture before client information enters an AI-supported workflow; bound delegated tasks with checkpoints and stop conditions; define a measurable value hypothesis; include implementation and change consequences; and make reusable deliverables governable.

These observations guide the model. They do not make one tool, cadence, method, deliverable, or approval rule universal.

Roles, interfaces, and accountability

Management consultant

Within assigned authority, the consultant:

• clarifies the question, decision, scope, assumptions, constraints, and evidence needs;
• plans and conducts authorized interviews, workshops, document review, observation, external research, and data analysis;
• separates verified facts, stakeholder perspectives, hypotheses, analytical judgments, and recommendations;
• maintains the engagement workplan, evidence log, issue log, decision log, and deliverable status for the assigned workstream;
• makes material uncertainty, limitations, contradictions, dependencies, and risks visible early;
• prepares options and a reasoned recommendation for the authorized client decision owner;
• supports implementation planning, handoff, measurement, and learning to the extent included in scope; and
• protects confidentiality, evidence integrity, version control, and human accountability regardless of the tools used.

Engagement lead

The engagement lead owns the integrated delivery within the consulting organization. Depending on [EMPLOYER POLICY], this role approves the workplan, allocates team capacity, reviews methods and deliverables, manages commercial and contractual boundaries, resolves cross-workstream dependencies, and escalates material risks. The engagement lead does not replace the client decision owner.

Client sponsor and decision owner

The authorized client sponsor confirms the business need, access, participation, constraints, success measures, governance route, and acceptance process. The client decision owner approves or rejects recommendations within the client’s authority model. Sponsorship, meeting attendance, or silence must not be treated as approval.

Client subject-matter and operational contributors

These contributors explain current work, supply authorized evidence, test factual accuracy, identify constraints, and review feasibility. Their statements are evidence from a perspective; they are not automatically the client’s final policy or decision.

Client control functions

Legal, privacy, compliance, security, finance, HR, procurement, records-management, risk, accessibility, and other control owners review matters within their assigned responsibilities. The consultant identifies the decision needed and provides evidence. The authorized control owner decides or approves according to [CLIENT RULE].

Consulting analysts and specialists

Team members perform assigned research, analysis, modeling, facilitation, writing, or technical work. The workstream owner gives a bounded brief, reviews the evidence and method, and remains responsible for integration. A specialist conclusion does not expand the approved engagement scope.

Required inputs and controlled records

Trigger-to-close engagement workflow

Use this sequence for a new engagement or a separately governed workstream. A small assignment may combine meetings or artifacts, but it should not skip the underlying decisions and controls.

1. Confirm intake, scope, and authority

The first task is to establish what the client is asking the team to support and who may decide. Review the signed scope and record the client’s requested outcome in decision terms rather than as a broad topic.

Actions:

• name the decision, decision owner, intended users, decision date, and reason the decision is needed;
• confirm deliverables, exclusions, commercial boundaries, locations, participants, and dependencies;
• distinguish the client’s desired outcome from the question the evidence must answer;
• identify matters reserved for executive, finance, HR, legal, privacy, security, procurement, risk, or operational approval;
• record assumptions that have not yet been verified; and
• stop and escalate any conflict between the request and the signed engagement scope.

Output: an accepted engagement or workstream brief with authority, scope, exclusions, and acceptance route.

Quality check: a new team member can explain the decision and boundaries without relying on an informal conversation.

2. Frame the problem and working hypotheses

Translate the broad issue into a structured set of questions that can be investigated. A useful frame defines the current state, desired state, gap, affected population, time horizon, constraints, decision criteria, and evidence needed.

Actions:

• write a neutral problem statement that does not assume the cause or answer;
• break the problem into distinct, collectively sufficient lines of inquiry suitable for this engagement;
• identify early hypotheses as testable propositions, not conclusions;
• define terms, units, populations, periods, comparison groups, and exclusions;
• establish decision criteria and materiality thresholds with the decision owner;
• identify what evidence would support, weaken, or disconfirm each hypothesis; and
• test whether the frame excludes a stakeholder, process, risk, or alternative explanation that could change the decision.

Output: a problem map, hypothesis register, key definitions, and decision criteria.

Quality check: every major line of inquiry connects to the decision; no branch exists merely because data is available.

3. Design discovery and evidence collection

Prepare a reproducible plan for learning how the organization actually works. Discovery can combine interviews, workshops, documents, process observation, operational data, surveys, external sources, prior analysis, and domain expertise when authorized and appropriate.

Actions:

• define each source family and what question it can credibly answer;
• set interview or workshop participant criteria and note important missing perspectives;
• prepare focused questions that distinguish practice, policy, perception, exception, and desired future state;
• establish inclusion, exclusion, date, geography, and quality rules for external research;
• confirm data classification, access, retention, recording, consent, citation, and distribution rules before collection;
• define how notes, transcripts, files, and extracts will be stored and quality-checked; and
• schedule interim synthesis so weak evidence or an incorrect frame is discovered early.

Output: an approved discovery plan, request list, interview or workshop guide, source protocol, and evidence log structure.

Quality check: the plan does not rely on one senior stakeholder, one convenient dataset, or one source family for a material conclusion.

4. Conduct discovery and maintain evidence integrity

During discovery, collect the smallest sufficient body of authorized evidence and preserve its context. The consultant should listen for disagreement, workarounds, exceptions, timing, authority, handoffs, and the difference between documented process and observed practice.

Ordered procedure:

1. Confirm the participant, purpose, confidentiality boundary, use of notes or recordings, and time available.
2. Ask for concrete recent examples before asking for general opinions.
3. Separate what the participant observed, inferred, prefers, or heard from someone else.
4. Test contradictions respectfully and record alternative explanations.
5. Capture source, date, process stage, owner, system, and limitation for material evidence.
6. Close with missing documents, follow-up questions, actions, owners, and permission to attribute or anonymize according to [CLIENT RULE].
7. Update the evidence, issue, assumption, and stakeholder logs promptly in [APPROVED SYSTEM].

Outputs may include interview notes, workshop decisions, process observations, document extracts, data inventories, research records, and a current-state evidence map.

Quality check: a factual claim can be traced to its source and context; a stakeholder opinion is not silently converted into an organizational fact.

5. Synthesize the current state and refine the frame

Bring the evidence together before running deeper analysis. Synthesis should show what is known, uncertain, disputed, missing, or outside scope.

Actions:

• code qualitative evidence consistently and preserve meaningful contrary cases;
• reconcile process documents with observed practice and system records;
• map work, decisions, information, delays, controls, handoffs, pain points, and exceptions;
• update the hypotheses and record why each was retained, revised, rejected, or deferred;
• distinguish root-cause candidates from symptoms and consequences;
• review preliminary facts with appropriate client contributors without asking them to approve the recommendation prematurely; and
• amend the plan when new evidence materially changes the question, while routing scope changes for approval.

Output: a validated current-state assessment, updated hypothesis register, evidence gaps, and analysis priorities.

Quality check: the synthesis includes counterevidence and does not imply that the most frequently voiced explanation is the cause.

6. Execute proportionate analysis

Select methods that match the decision, evidence quality, and risk. Analysis may include descriptive comparison, segmentation, trend or variance analysis, process analysis, financial modeling, scenario analysis, forecasting, qualitative coding, option assessment, risk analysis, or other domain-specific methods approved for the engagement.

Ordered procedure:

1. Freeze the analysis question, population, period, variables, definitions, exclusions, source versions, and acceptance criteria.
2. Check completeness, duplicates, missingness, outliers, units, joins, transformations, and known collection bias.
3. Run the simplest method that can answer the question credibly before adding complexity.
4. Reconcile key totals and transformations to the source or governed semantic definition.
5. Test sensitivity to material assumptions, alternative definitions, and plausible scenarios.
6. Search for disconfirming evidence and explanations that produce a similar pattern.
7. Review generated code, formulas, summaries, classifications, and visualizations before use.
8. Record method, assumptions, checks, limitations, reviewer, and reproducible output location in an analysis note.

Output: analysis files, an audit note, verified exhibits, limitations, and implications linked to the decision criteria.

Quality check: the analysis supports the stated conclusion but does not claim causation, precision, prevalence, or forecast certainty beyond the method and evidence.

7. Develop options and a recommendation

Convert findings into realistic choices. An option is not credible unless the client can understand what it changes, what it requires, and what trade-offs it creates.

Actions:

• define the baseline or no-change path and at least the realistic alternatives warranted by the evidence;
• assess options against the approved criteria, including value, cost, timing, feasibility, risk, capability, stakeholder impact, controls, and reversibility where relevant;
• separate observed facts, assumptions, modeled effects, stakeholder preferences, and consultant judgment;
• state dependencies and conditions that could change the ranking;
• identify implementation, operating-model, data, technology, communication, training, governance, and measurement implications;
• develop a clear point of view while preserving unresolved dissent and uncertainty; and
• route legal, regulatory, financial, security, employment, procurement, accessibility, or other specialist conclusions to the authorized owner.

Output: option comparison, risk and dependency view, value hypothesis, preferred recommendation, and conditions for decision.

Quality check: the recommendation follows from the evidence and criteria; it is not simply the client’s initial preference or the most sophisticated solution.

8. Build and quality-review the decision package

The main deliverable should help an authorized audience decide. Its executive layer should be concise; its evidence layer should preserve the material sources, assumptions, calculations, limitations, and update rules needed for challenge or reuse.

Actions:

• lead with the decision, answer, rationale, material evidence, risks, and requested action;
• show only charts, tables, process maps, models, or examples that advance the decision;
• label modeled, illustrative, estimated, or fictional values clearly;
• reconcile every number, chart, and statement across the memo, presentation, model, dashboard, and appendix;
• check names, dates, units, denominators, version, confidentiality marking, accessibility, and distribution;
• remove unsupported certainty, unexplained jargon, decorative analysis, hidden assumptions, and internal drafting notes;
• complete independent fact, method, logic, public-copy, and senior review according to [EMPLOYER POLICY]; and
• retain the approved version and evidence cutoff in [APPROVED SYSTEM].

Output: a decision memo, executive briefing, supporting evidence layer, and review record appropriate to the engagement.

Quality check: the intended decision owner can explain the recommendation, alternatives, risks, evidence strength, and required decision after reading the package.

9. Facilitate the decision and record conditions

The consultant presents the recommendation and supports a decision; the consultant does not manufacture approval.

Ordered procedure:

1. Confirm the meeting purpose, participants, authority, time, pre-read status, and decision requested.
2. State the question and recommendation before presenting detail.
3. Explain the evidence, important uncertainty, alternatives, trade-offs, and implementation conditions.
4. Invite challenge and distinguish requests for clarification from objections that change the decision.
5. Record the authorized decision, conditions, dissent, deferred items, owners, dates, and follow-up evidence.
6. Do not treat tentative support, meeting attendance, or absence of objection as approval.
7. Circulate the decision record through [APPROVED SYSTEM] according to [CLIENT RULE].

Output: a decision record and authorized next-step list.

Quality check: the recorded decision names the actual authority owner and any conditions that must be satisfied before action.

10. Prepare implementation handoff and measurement

Where implementation is in scope, translate the recommendation into controlled action. Where it is not, provide only the agreed handoff detail and avoid implying delivery responsibility.

The handoff should cover:

• outcome and value hypothesis;
• baseline, leading and lagging measures, owner, source, and review cadence;
• workstreams, sequence, dependencies, milestones, and decision gates;
• accountable owners and contributors;
• affected stakeholders, communications, training, adoption, and support;
• data, technology, process, policy, budget, capacity, vendor, and control prerequisites;
• risks, triggers, mitigations, contingency, and escalation routes;
• pilot, test, acceptance, stop, adjust, and scale rules where applicable;
• deliverable owners, update instructions, access, retention, and maintenance; and
• items explicitly outside the consultant’s responsibility.

Output: an implementation roadmap or handoff note, action register, measurement plan, and risk/decision schedule.

Quality check: each material action has an owner, due date, dependency, expected evidence, and governing approval.

11. Close the workstream and capture learning

Close only after accepted deliverables, decisions, actions, access, records, and retention requirements are reconciled.

Ordered procedure:

1. Confirm delivery and acceptance against the agreed criteria.
2. Transfer approved files, models, dashboards, decision records, action logs, and update instructions to the named owners.
3. Close, transfer, or explicitly retain open risks, issues, assumptions, and decisions.
4. Revoke or adjust temporary access and dispose of data according to [EMPLOYER POLICY], [CLIENT RULE], and the engagement agreement.
5. Compare the workplan with actual delivery and record material variance without inventing outcome claims.
6. Capture reusable method learning without copying confidential client content into a general knowledge base.
7. Agree any authorized post-engagement measurement or review point.

Output: closure record, ownership handoff, open-item register, access/retention confirmation, and learning note.

Quality check: nothing material remains ownerless, hidden in a personal workspace, or presented as complete when acceptance is unresolved.

Operating cadence

Cadence depends on engagement size, risk, duration, client governance, and decision timing. Replace the examples below with [ENGAGEMENT-SPECIFIC] frequencies.

Daily rhythm

• Review immediate priorities, deadlines, dependencies, evidence requests, and decision-owner needs.
• Update workplan status, actions, new assumptions, issues, and access constraints.
• Quality-check new interview notes, data extracts, calculations, generated summaries, and draft exhibits before reuse.
• Escalate confidentiality, authorization, safety, factual, deadline, or scope concerns without waiting for the next scheduled meeting.
• Protect focus time for analysis and synthesis rather than allowing meetings to replace work.

Weekly rhythm

• Hold a workstream review covering completed evidence, open gaps, analysis results, next decisions, risks, and capacity.
• Reconcile the client action/decision log and close items only with evidence.
• Review hypotheses, contradictions, scope pressure, stakeholder coverage, data quality, and version control.
• Conduct an engagement-lead quality review of material analysis and upcoming client deliverables.
• Share a concise client status update that separates progress, decisions needed, risks, next work, and owner actions.

Monthly or stage-gate rhythm

For engagements long enough to require it:

• review delivery against scope, milestones, budget or effort, quality, risk, and stakeholder expectations;
• check whether the decision question, value hypothesis, baseline, and measures remain valid;
• review access, confidentiality, retention, AI-use, and control exceptions;
• reassess dependencies, implementation readiness, adoption risks, and handoff ownership; and
• approve, revise, pause, or close the next stage according to [CLIENT RULE] and [EMPLOYER POLICY].

Event-driven rhythm

Run an immediate bounded review when any of these events occurs:

• material new evidence contradicts the current conclusion;
• the client changes the decision, scope, deadline, users, or success criteria;
• an executive requests a commitment outside the approved scope;
• a dataset, calculation, model, or source fails a material quality check;
• sensitive information is shared through an unapproved route or access appears excessive;
• an AI-supported output or connected action is inaccurate, unsafe, unauthorized, or ambiguous;
• a stakeholder dispute reveals unclear authority or an unrepresented affected group;
• implementation conditions become infeasible; or
• a legal, regulatory, financial, employment, security, privacy, accessibility, or safety issue requires an authorized specialist.

Decision rights and approval boundaries

If authority is uncertain, treat the decision as reserved and escalate. Never infer approval from seniority, urgency, meeting attendance, copied email recipients, or the output of a tool.

Handoffs and escalation

Decision-ready escalation record

An escalation should contain:

• the decision or intervention required;
• the relevant engagement scope, policy, or authority boundary;
• verified facts, source dates, and material uncertainty;
• impact of action, delay, or no action;
• realistic options and trade-offs;
• the consultant’s recommendation, when appropriate;
• the named authority owner;
• the deadline and reason;
• an interim control that limits risk while waiting; and
• the final decision, conditions, action owner, due date, and closure evidence.

Escalate immediately when there is suspected unauthorized disclosure, unsafe external action, material analytical error in a released deliverable, instruction to misstate evidence, unclear authority for a consequential decision, or a conflict between the request and the approved engagement scope.

Handoff quality standard

A receiving owner should not need to reconstruct the analysis or guess what to do. The handoff should identify the purpose, current status, accepted evidence, decisions already made, open questions, assumptions, dependencies, risks, next action, owner, due date, access, record location, and review cadence.

The consultant verifies receipt and understanding. Sending a file does not by itself complete a handoff.

Quality checks and measures

Problem and evidence quality

• The decision, owner, deadline, criteria, scope, exclusions, and definitions are explicit.
• The source plan covers the relevant perspectives and does not rely on a single convenient voice.
• Material claims are traceable to dated evidence with provenance and limitations.
• Contrary evidence, missing data, uncertainty, and unresolved disagreement remain visible.
• The evidence geography, period, population, and decision context match the claim.

Analysis quality

• Variables, units, populations, periods, exclusions, joins, transformations, and assumptions are documented.
• Key totals and outputs reconcile to the source or approved definition.
• Sensitivity and plausible alternative explanations are tested where they could change the decision.
• Qualitative themes retain meaningful counterexamples and are not converted into unsupported prevalence.
• Generated code, formulas, classifications, summaries, and visuals receive human review.

Recommendation quality

• Options are real, distinct, and assessed against agreed criteria.
• The recommendation distinguishes fact, inference, modeled effect, assumption, preference, and judgment.
• Costs, risks, dependencies, affected stakeholders, controls, and feasibility are visible.
• The value hypothesis has a baseline, measure, source, owner, cadence, and stop/adjust/scale rule.
• Implementation and handoff responsibilities match the approved engagement scope.

Delivery quality

• The executive answer is understandable without reading every appendix.
• Numbers, charts, narrative, actions, and appendices agree.
• Confidentiality, distribution, version, accessibility, citation, and review requirements are satisfied.
• The client decision and conditions are recorded by the correct authority.
• Reusable models or dashboards have an owner, update method, assumptions, access, and maintenance rule.

Useful operating measures

Use measures to improve the work, not to manufacture certainty. Possible measures include:

• percentage of material claims with complete source and review fields;
• evidence-request aging and unresolved critical gaps;
• analysis defects found before versus after client delivery;
• decisions made, deferred, or escalated with named owners;
• action completion supported by closure evidence;
• deliverables accepted against agreed criteria;
• forecast or business-case variance classified by cause when measurement is in scope;
• stakeholder coverage against the approved discovery plan; and
• AI-assisted outputs accepted, corrected, or rejected after review.

Thresholds, targets, retention, and reporting routes are [EMPLOYER POLICY] or [CLIENT RULE].

Tool categories and controlled use

Use tool categories rather than assuming a universal product stack:

• spreadsheets and financial or operational models;
• presentation, document-authoring, and diagramming tools;
• survey, interview, workshop, and qualitative-analysis tools;
• data querying, statistical, scripting, and notebook environments;
• BI, dashboard, metric, and visualization platforms;
• process, workflow, project, risk, CRM, collaboration, and enterprise systems;
• domain-specific finance, technology, cloud, security, operations, or regulatory systems; and
• approved AI-assisted research, analysis, drafting, evaluation, and workflow tools.

For every material tool use, identify the approved environment, source of record, access, purpose, version, data classification, permitted action, human reviewer, export or sharing rule, and retention requirement. A convenient personal tool is not an [APPROVED SYSTEM].

Confidentiality and AI-assisted work boundaries

Before use

• Classify the information and confirm the business purpose and authorization.
• Minimize the input and remove client or personal identifiers when they are not necessary.
• Confirm approved account, model or service, region, connector, retention, training-use, logging, and access configuration under [EMPLOYER POLICY] and [CLIENT RULE].
• Define the permitted task, prohibited actions, output owner, tests, checkpoints, stop conditions, and escalation route.
• Use fictional or sanitized context for method development when real data is not authorized.

Appropriate bounded assistance when approved

• generate research questions or interview prompts for human review;
• summarize authorized material while preserving source links and checking against the original;
• help classify qualitative evidence using a reviewed codebook;
• draft or explain formulas, queries, or scripts that a qualified person tests;
• compare scenarios using supplied assumptions;
• identify anomalies or contradictions for investigation;
• improve structure, clarity, accessibility, or consistency of a draft; and
• create a provisional work product that remains subject to consultant review.

Prohibited without explicit authority and controls

• upload restricted client information to an unapproved service;
• infer or enrich personal, sensitive, confidential, or protected information beyond the authorized purpose;
• fabricate interviews, sources, facts, citations, calculations, client views, or model output;
• let an agent change a client system, send a message, publish a file, accept terms, or create a commitment without the approved action boundary;
• present generated content as verified evidence without source reconciliation;
• allow a tool to make a legal, employment, safety, financial, security, privacy, regulatory, procurement, or executive decision; or
• use client information to build a general reusable asset when the engagement terms do not authorize that use.

Human validation record

For consequential AI-supported work, record the task, tool and configuration category, data boundary, source material, output, reviewer, tests, corrections, rejected elements, final use, and retained evidence in [APPROVED SYSTEM]. The consultant remains accountable for the released claim or action.

Exception handling

New evidence changes the conclusion

Pause the affected conclusion, trace every dependent claim and exhibit, update the analysis, inform the engagement lead and client owner, and reissue only through approved version control. Do not quietly edit a final deliverable whose decision context has changed.

The client requests work outside scope

Clarify the request, value, urgency, effort, risk, dependencies, and affected deliverables. Continue only the existing authorized work until the engagement lead and client authority approve a change under [EMPLOYER POLICY] and the contract.

Evidence is insufficient

State what is missing, why it matters, what alternatives exist, and whether the decision can be deferred, narrowed, tested through a pilot, or made with an explicit risk. Do not convert absence of evidence into a convenient assumption.

Stakeholders disagree

Record the points of agreement, competing claims, evidence, interests, authority, and decision needed. Facilitate the issue; do not erase dissent to make the presentation cleaner.

A released analysis contains a material error

Stop reuse, notify the engagement lead, determine affected outputs and decisions, preserve the evidence, correct through approved version control, inform affected client owners, and record the response according to [EMPLOYER POLICY] and [CLIENT RULE].

An AI or connected-tool action is ambiguous

Do not repeat the action. Reconcile read-only records, logs, versions, and client-system state. Escalate if the state cannot be established without another mutation.

Quick-reference checklist

Before work begins

• Confirm the decision, owner, deadline, scope, exclusions, acceptance route, and authority map.
• Approve the research, data, confidentiality, access, tool, retention, and distribution rules.
• Define the evidence needed to support, weaken, or reject the starting hypotheses.
• Establish the workplan, evidence log, issue/risk log, decision/action log, and version control.

Before analysis is accepted

• Verify population, period, definitions, source versions, transformations, reconciliation, and missingness.
• Test material assumptions, contrary evidence, and alternative explanations.
• Review generated code, formulas, classifications, summaries, and visualizations.
• Record method, checks, limitations, reviewer, and reproducible output location.

Before a recommendation is delivered

• Separate facts, inference, modeled effects, assumptions, preferences, and judgment.
• Compare realistic options using approved criteria.
• Make costs, risks, dependencies, affected stakeholders, controls, and feasibility visible.
• Reconcile the narrative, numbers, exhibits, actions, and appendices.
• Confirm confidentiality, accessibility, distribution, citations, version, and approvals.

Before closure

• Record the authorized decision and every condition.
• Transfer deliverables, records, open items, update instructions, and ownership.
• Confirm measurement, review cadence, stop/adjust/scale rules, and escalation paths.
• Reconcile access, retention, return, and disposal.
• Capture learning without moving confidential client content into an unauthorized reusable asset.