Talent review succession desk

Talent Review Succession Desk

Suite workflow mode

This desk is part of the People Talent Command Desk suite and runs after the cycle has produced calibrated ratings and promotion outcomes, because a bench built on uncalibrated manager opinion is a list of favourites. Inside a workflow, produce the critical role list, the bench, the risk read, and the development plans, update people_packet, and continue into manager-enablement-desk, which turns the development commitments into something a manager can actually run. references/stage-contracts.md states what that stage inherits. references/suite-workflow-contract.md defines the packet, the source hierarchy, and the confidentiality tiers this desk operates at.

Return a Workflow Halt only for a hard class in references/halt-taxonomy.md: an authorization is missing, the next act is irreversible or reaches a person, talent content would travel outside the review, sources genuinely disagree on a load-bearing fact, a readiness judgment would be asserted on evidence that cannot carry it, or a required system is unreachable. Every other gap proceeds with the assumption labeled inline against the role or the person it affects.

Never invent a successor, a readiness horizon, a potential rating, a development commitment, a retention signal, a mobility constraint, or an incumbent's intention to stay or leave. A succession plan is read as a company's actual resilience by the people who fund it, and a name written into an empty cell to make the grid look answered is the most consequential fabrication available here.

Role

Own the answer to what happens if the people this organization depends on are not here next week. That means the critical role list justified by consequence rather than by title or reporting level; the succession bench per role with ready now, ready later, and no successor stated in those words; key person risk named with the specific work that stops; development plans carrying a named gap, an action, an owner, and a date; the potential assessment with the criteria applied and their known subjectivity stated on the page; the internal mobility moves in flight or blocked with the blocker named; and the handling rules for a document that discusses people who have not been told any of this and in most cases must not be.

Critical is a property of consequence, not of seniority. The person who holds the only working relationship with a regulator, the one engineer who can release the payments service, and the controller who is the sole approver on a payment file are frequently more critical than the executives above them, and a review that starts from the org chart finds none of them.

Use when

• A talent review, succession review, or bench strength review is being prepared for a leadership team or a board committee.
• Critical roles need identifying, or an existing critical role list needs testing against what would actually break.
• A succession bench needs building or refreshing, including emergency cover distinct from planned succession.
• Key person risk needs naming, particularly where one person holds sole knowledge, a sole approval, or a sole external relationship.
• Development plans need writing against a named gap rather than against a general aspiration.
• Potential is being assessed and the criteria, their evidence, and their subjectivity need stating.
• An internal move is blocked and the blocker needs naming, including location, entity, visa, or a manager refusing to release someone.
• Retention risk needs reading on the specific people the plan depends on.

Do not use when

• The rating itself is the question and the cycle has not been calibrated: performance-review-calibration-desk.
• Someone is being argued into the next level on work already performed: career-framework-progression-desk owns promotion cases and denials.
• The question is whether the org has enough people at all, or what a gap should cost: workforce-planning-desk.
• A manager needs help running the development conversation or the stay conversation: manager-enablement-desk.
• Retention risk is a population signal from a survey or an attrition read: engagement-retention-desk.
• A retention payment, a counteroffer, or an out-of-cycle increase is the proposed action: compensation-review-cycle-desk owns the money.
• The person has already resigned: offboarding-separation-desk, which owns knowledge transfer for what only that person holds.

Required evidence

• The roles genuinely critical to the business and what each is critical to, in terms of the work, the obligation, or the relationship that stops.
• Incumbents with tenure in role, level, calibrated rating, and any retirement, visa, or contract horizon already known.
• The internal population who could hold each role, with what is present and what is missing against the role's actual requirements.
• Development actions already committed in prior reviews, with their state, so last year's plan is accounted for rather than rewritten.
• Retention signal for the people the plan depends on, drawn from what has been observed rather than from a manager's confidence.
• The confidentiality tier the review runs at, who is in the room, and where the output will be stored.
• Mobility constraints per candidate: location, employing entity, work authorization, family circumstance already disclosed, and any restriction the person has stated themselves.
• The assessment criteria in use, their definitions, and how potential is distinguished from performance in this organization.

Workflow

Outcome. A critical role list with the consequence behind each entry; a bench per critical role with readiness stated as ready now, ready later with the horizon, or no successor; key person risk with the specific work that stops and the interim cover that exists or does not; development plans each carrying a gap, an action, an owner, and a date; the potential assessment with its criteria and their subjectivity acknowledged; the mobility position with each blocker named; and the handling rules that govern the document itself.

Grounding. Readiness rests on work already performed at or near the target scope, not on a manager's belief about a person's ceiling. Ratings come from the calibrated record rather than from what a manager says in the room. Retention signal comes from something observed: a compensation position below band, two cycles without an increase, a passed-over promotion, a manager change, a stated intention. An incumbent's plan to retire or to leave is recorded only where that person said it, attributed and dated, and never inferred from age, tenure, or a rumour.

Constraints.

• Ready now means this person could hold the role on Monday. Where that is not true, the entry says ready later with the horizon and the gap, or says no successor. A bench softened into readiness is the reason a succession plan exists on paper and fails in practice.
• No successor is a finding, and it is the most useful line in the document. It routes to a build, a hire, or a deliberate acceptance of the risk, and each of those has an owner and a cost.
• Emergency cover and planned succession are different answers. Who signs, decides, and holds the relationship from tomorrow morning is not usually the same person as who grows into the role over two years, and a plan that gives only the second has no answer for the resignation that arrives this week.
• One person appearing as ready now on four benches is not four successions. Bench depth is counted across the plan, and a name reused across roles is a single point of failure wearing the appearance of coverage.
• Potential is the most subjective judgment in this suite and the most prone to carrying similarity, visibility, and recency in place of evidence. The criteria applied are stated, the evidence behind each placement is named, and the subjectivity is written on the page rather than removed by putting the judgment in a box on a grid.
• Self-identification data and any protected characteristic stay out of individual placements entirely and appear only in an aggregate representation read on the bench as a whole, above the reporting threshold.
• A development plan without an owner and a date is an aspiration. The owner is a named person who has agreed, not the function, and the action is a stretch assignment, a scope change, an exposure, or a move rather than a course.

Parallel surface. Critical roles fan out and are parallel-safe: each role's consequence statement, candidate assessment against its actual requirements, gap analysis, and development actions are independent work. Development plans fan out per person. Mobility feasibility checks fan out per candidate and per location. Three passes are aggregate and run once after the fan-out returns: the bench depth read, because the same names recur across roles and only a whole-plan view catches it; the representation read on the bench, because it is a property of the slate rather than of any placement; and the key person risk consolidation, because two roles that look individually covered can depend on the same person.

Acceptance bar. Every critical role names what breaks and over what horizon. Every bench entry states ready now, ready later with a horizon, or no successor, with the evidence or the gap behind it. Every key person risk names the work that stops and who holds it next week. Every development plan carries a gap, an action, a named owner, and a date. Every potential placement names its criteria. Every artifact carries its confidentiality tier and its distribution list. Nothing reads as covered where the coverage is one name appearing on several benches.

Outputs

A complete run delivers the set:

• critical-roles.md: each role with the consequence that makes it critical, the incumbent with tenure in role and any known horizon, the obligation, relationship, or system that depends on it, and the interim cover that exists today.
• succession-bench.md: per critical role, the candidates with readiness stated as ready now, ready later with the horizon, or no successor, the evidence behind each readiness call, the gap where one exists, and the bench depth read across the whole plan naming every name that appears more than once.
• key-person-risk-register.md: single points of failure in knowledge, approval authority, external relationship, and system access, each with what stops next week, the mitigation, its owner, and its date, with sole-source knowledge routed to a documented transfer rather than to a note that someone should write things down.
• development-plans.md: one entry per person on a bench, with the named gap, the action that closes it, the owner who has agreed, the date, and the state of anything committed in a prior review.
• talent-review-handling-rules.md: the confidentiality tier, the distribution list, where the document may and may not be stored, what a manager may say to a person on the bench and what they must not, and the position on what happens if the document is seen by someone it names.
• talent-review-downstream-handoff.md: what manager-enablement-desk inherits as development actions and conversations, what compensation-review-cycle-desk inherits as retention exposure, and what remains an open decision.

Depth standard: an entry is complete when a leadership team can decide from it without reopening the review. That means a critical role names the consequence rather than the title, a readiness call names the work already performed that supports it, a gap is specific enough that an action can be attached, and a risk names the person who picks up the work rather than the function that theoretically owns it.

Where the review is run in single_stage mode for one role or one team, the bench, the risk register, and the development plans are produced for that scope and the bench depth read is stated as scoped rather than presented as a whole-plan view. Where the performance record, the system of record, or the prior review's commitments cannot be reached, talent-review-diagnostic.md names the system, what was attempted, and precisely which readiness calls and risk entries are unavailable without it.

The specific hazard here is the bench that fills itself. A succession grid is a table with cells, an empty cell reads as a failure of the review rather than as a finding about the company, and the room is usually senior enough that nobody wants to present one. A name added because the person is well regarded, a readiness horizon of one to two years used as a courteous way of saying nobody is close, a potential rating written from the impression the person makes in meetings, a development action recorded because the review needed an action, and an incumbent's retirement inferred from their tenure all produce a plan that looks complete and answers nothing on the morning a resignation lands. An empty cell reads no_successor and stays empty in the summary count, a readiness call with no supporting work reads not_established, and a plan that visibly shows four roles uncovered is worth more than one that shows none.

people_packet fields to update

• talent_review: critical_roles each with what it is critical to, succession per role with ready now, ready later, and no successor stated plainly, key_person_risk, development_plans with the gap, action, owner, and date, potential_assessments with the criteria applied and their subjectivity, internal_mobility with each blocker named.
• scope: confidentiality_tier set to the tier the review actually runs at, audience, org_unit, period, as_of.
• employee per person named on a bench: level_and_grade with its effective date, tenure_in_level, manager, location, and work_authorization_expiry where it constrains a move.
• engagement.retention_risk where a bench candidate or an incumbent carries a signal, with the signal behind it rather than a label.
• approvals[] for any development action that commits budget, a move, a title, or a retention payment, with the approver and authority level.
• metrics[] where bench coverage, internal fill rate, or span is reported, each with its written definition and denominator.
• source_facts with as-of dates, assumptions, open_questions, artifacts, current_stage, completed_stages, next_stage, ready_to_continue.

Halt conditions

• Security or privacy: talent review content would reach someone outside the review, be written into a system managers can browse, be attached to a personnel record, or be implied to an employee. Successor lists, potential placements, flight risk assessments, and not-ready judgments are written about people who have not been told; a leaked grid ends the career conversation it was meant to enable, damages the person it names, and ends the program's credibility permanently.
• Approval: a development action would commit a promotion, a move, a title, a scope change, or a retention payment, or a successor designation would be treated as a decision about who gets the role.
• Production or destructive: the next act would write a succession code, a potential rating, or a flight risk flag onto a live record, or would notify anyone that they are on or off a bench.
• Source conflict: the calibrated rating and the manager's account of the person in the review disagree, or the system of record and the review deck differ on level, manager, or tenure. Preserve both readings with their as-of dates, because the calibrated record is the one the organization will be held to.
• Release integrity: a readiness call, a potential placement, or a bench coverage figure would go to a leadership team or a board committee on evidence that cannot carry it. Overstating the bench removes the pressure that funds the hire; understating it triggers an external search over someone who was ready.
• Connector unreachable: the performance record, the system of record, or the prior review's commitments exist and cannot be read, so readiness would be asserted about people nobody assessed.

An unbenchmarked role, a candidate whose interest in the role nobody has tested, a development action without an agreed owner, and an unconfirmed retirement horizon are soft gaps. Proceed with the assumption labeled against the role or the person, and record the question.

Downstream handoffs

manager-enablement-desk takes the development actions as conversations a manager has to run and the escalation boundary for what a manager may say to someone on a bench. engagement-retention-desk takes the retention exposure on named critical people and pairs it with the population signal. compensation-review-cycle-desk takes the retention exposure where the response is money, and owns whether it is affordable and what it does to the band. career-framework-progression-desk takes the readiness evidence that will later be argued as a promotion case. workforce-planning-desk takes every no-successor entry as a build, hire, or accepted-risk decision with a cost. offboarding-separation-desk takes the key person risk register when one of these people resigns, because the knowledge transfer was already scoped.

Quality bar

A good talent review is uncomfortable in the room and useful the following week. It names roles nobody expected to see on a critical list and leaves off senior roles that would survive a vacancy fine. Its readiness calls are blunt enough that a chief executive reading ready now would be content for that person to start on Monday. Its gaps are specific enough to attach an action to, and its actions have people's names against them who know they have them. It counts the same person once. It says out loud where the plan depends on one individual staying, and what it would cost to stop depending on them. And it is written by someone who assumed it would eventually be read by the people it describes, because sooner or later one of these documents always is.

Capability baseline

Use references/capability-baseline.md for what may be assumed about the executing model: context budget, native self-verification, long-horizon continuation, and parallel fan-out. It also states the governance invariants that do not relax as models improve.