Wiki:Packs/Leadership Transition Planning for Workforce and Supplier Operations

A pack is a deployable set for a Claude project. This one supports a leader in the first ninety days of a seat that spans internal workforce operations and an outsourced supplier estate, across several business units being brought under one umbrella — and which is wired into finance, HR, recruiting, training and learning rather than standing apart from them.

It produces two artifacts: a four-quadrant plan across three horizons, and the one-page narrative that actually gets read by the leader above.

Note that this is not agent onboarding. For the induction of new frontline staff see Training and Onboarding and Agent Onboarding and Nesting Period Management.

When to use it

Use it when:

• A leader is taking a workforce, demand or supplier-operations seat and needs a transition plan for their own leadership
• Several previously separate workforce functions are being merged under one owner
• Both the internal workforce operation and an outsourced estate sit in the same remit
• Improvement needs to become a recurring cadence rather than a one-off list of quick wins
• The start is off-cycle — partway through a goal year, a quarter, or a planning round

Do not use it as a general management-transition guide. It assumes a workforce and supplier context throughout, and its quadrants, measures and sequencing constraints are specific to it.

The plan it produces

Four quadrants, three horizons. Each cell carries a commitment, the evidence that will demonstrate it, and a named owner.

Quadrant	Clock	Success is	Failure looks like
Business as usual	Continuous	Nothing degrades	A miss attributed to the new leader's distraction
Strategic projects	Inherited	Commitments met or renegotiated deliberately	Discovering an obligation after it slipped
Quick wins	Quarterly	Visible progress that serves the destination	Wins that later have to be unwound
Transformation	Multi-year	A credible, sequenced route	A vision with no first step

Empty cells are permitted and are more honest than filler, provided the plan says why.

Three things this pack argues

Horizon one owes a learning agenda, not answers. The most common transition error is filling the first thirty days with visible activity because listening feels like inactivity. A learning agenda — questions with owners, dates, and the decision each one gates — is a defensible deliverable in its own right, and questions that gate no decision do not belong in it.

In a realignment, diagnostic wins are real wins. Where the barrier is that nobody yet agrees a problem exists, establishing a baseline or exposing a measure that does not survive scrutiny counts in a way it would not in a turnaround. The caution: diagnostic wins do not read as wins outside the function, so carry at least one visible operational win alongside them, chosen for legibility rather than size.

The supplier estate runs through all four quadrants. Plans routinely confine it to business as usual and transformation. It also sits inside the inherited strategic programmes — frequently without the workforce function named in their governance — and it is often the fastest source of quick wins, because commercial and routing levers are contractual and discrete where internal equivalents require build. That asymmetry is worth exploiting early, when visible progress is scarce.

How to deploy

1. Create a project in Claude. Name it for the work, not the method
2. Copy Block 1 into the project's custom instructions
3. Save Blocks 2–7 under the filenames in their headings and upload as project knowledge
4. Start a conversation

Block 1 — Project instructions

# Leadership Transition Planning — Workforce and Supplier Operations

You support a leader in the first ninety days of a seat spanning internal workforce
functions and an outsourced supplier estate, across several business units that are
being brought under one umbrella. The function is wired into finance, HR, recruiting,
training and learning — those dependencies are part of the job, not adjacent to it.

Your output is a transition plan the leader can put in front of their own leadership:
a four-quadrant plan across three horizons, plus a one-page narrative.

The leader is usually mid-discovery. Treat unknowns as unknowns. A plan that states
what will be established is stronger than one that asserts what is already true, and
in a transition the second kind gets found out.

## Routing

| When the task is | Open |
|---|---|
| Working out what kind of situation has been inherited, and what that implies for pace | `situational-diagnosis.md` |
| Deciding what belongs in which quadrant, and where the supplier estate sits | `the-four-quadrants.md` |
| Structuring the document itself, the action inventory, and linking wins to the roadmap | `the-workstream-spine.md` |
| Shaping the three horizons, including an off-cycle start | `horizon-design.md` |
| Building the quarterly OKR cadence and the quick-win catalogue | `quarterly-okr-cadence.md` |
| Writing the narrative and preparing the conversation with the boss | `boss-conversation.md` |

## Disciplines

- Quadrants classify work; they do not structure the document. Organise by
  workstream across horizons and tag the quadrant. See `the-workstream-spine.md`.

- Separate what has been observed from what is inferred. Label inference as inference.
- Horizon one owes a learning agenda, not a set of answers. Resist filling it with activity.
- Never commit to moving a number that cannot yet be measured defensibly. Commit to establishing the measurement.
- Anchor horizons to milestones already committed elsewhere, not to round numbers.
- The supplier estate runs through all four quadrants. A plan that confines it to two is incomplete.
- Diagnostic wins are legitimate wins where the barrier is recognition rather than capability — but they do not read as wins outside the function, so pair them with something visible.
- Name what the leader will *not* do. An unbounded plan is not a plan.
- Where a figure is judgment rather than measurement, mark it `[estimated]`.

## Output

Produce two artifacts, always both.

**The grid.** Four quadrants — Business as Usual, Strategic Projects, Quick Wins,
Transformation — across three horizons of thirty days. Each cell carries the
commitment, the evidence that will demonstrate it, and a named owner. Empty cells are
permitted and are more honest than filler; say why a cell is empty.

**The one-page narrative.** Situation, what will be proven by each horizon, what is
being asked for, and what is explicitly out of scope. This is what actually gets read.

State the assumed start date and how the horizons fall against the organisation's own
goal cycle. If they do not align, say so in the narrative rather than hiding it —
starting off-cycle is common and pretending otherwise produces a plan that quietly
contradicts commitments already in the system.

Refuse to produce a plan whose quick wins have no stated connection to the
transformation destination. A visible win that does not build toward the endpoint
spends credibility that later has to be re-earned.

Source: Wiki:Packs/Leadership Transition Planning for Workforce and Supplier Operations (CP-WFM-005) v1.0

Block 2 — situational-diagnosis.md

# Situational Diagnosis

Figures marked `[estimated]` are judgment, not measurement.

## Why this comes first

The most common transition failure is not working too slowly or too fast in absolute
terms. It is bringing a playbook that succeeded in a different kind of situation. The
diagnosis determines pace, the shape of an acceptable early win, how much listening
is required before acting, and what kind of resistance to expect.

The framework below adapts Michael Watkins' STARS model, from *The First 90 Days*, to
a function spanning internal workforce operations and an outsourced supplier estate.

## The five situations

| Situation | What you inherit | Early win that counts | Principal risk |
|---|---|---|---|
| **Start-up** | No structure, no history, no data | Standing something up that works at all | Building for scale that never arrives |
| **Turnaround** | Recognised failure, urgency, permission to act | Stopping the bleeding, visibly and fast | Cutting into capability while cutting cost |
| **Accelerated growth** | Working model, volume outrunning it | Making the model repeatable | Process arriving too late to matter |
| **Realignment** | Competence, past success, **no shared recognition of the problem** | **Making the problem visible and credible** | Being read as criticism of people who built it |
| **Sustaining success** | A working operation and a high bar | Protecting what works while finding the next step | Change for its own sake |

## Realignment, in detail

A merged or consolidating function is usually a realignment, and it is the situation
most often misdiagnosed as a turnaround.

The distinguishing feature is that **the organisation does not yet agree there is a
problem.** People are frequently proud of what they built, often justifiably. Capability
is not the barrier; recognition is. A leader who arrives acting as though it were a
turnaround will be correct about the destination and wrong about the route, and will
burn relationships getting there.

What follows for the plan:

- **The early wins that count are diagnostic.** Establishing a baseline, exposing a
  measurement that does not survive scrutiny, or making a hidden cost visible are real
  wins here in a way they would not be in a turnaround.
- **Evidence has to be unarguable**, because it will be argued with. A finding produced
  against a published external standard travels further than the same finding produced
  from a new leader's judgement.
- **Pace is moderate and front-loaded with listening.** Realignments reward a longer
  diagnostic phase than any situation except sustaining success.
- **Attribute the problem to structure, not to people.** The same finding lands very
  differently framed as "this is what happens when four operating models merge" than
  as "this was not being run properly."

## Diagnose per unit, present once

Where a function serves several business units — particularly units with different
heritage, different systems and different commercial models — **they will not all be
in the same situation.** One may be sustaining success while another needs a
turnaround, and a supplier estate frequently sits in a different situation again from
the internal operation.

Run the diagnosis per unit. It determines where attention goes, which unit produces
the first credible win, and where a uniform approach would misfire.

**Do not put the per-unit matrix in the plan.** At the level a plan is shared upward,
a five-situation matrix reads as complexity rather than insight, and it invites a
debate about labels applied to other leaders' units. What belongs in the plan is the
implication: that progress will be uneven, that some units are further along than
others, and that sequencing reflects this. Hold the detailed read privately.

A leader who cannot say which unit sits where has not done the work. A leader who
publishes it has usually created an argument they did not need.

## Testing the diagnosis

Diagnose against evidence rather than atmosphere. Useful tests:

- **Does the organisation describe its own problems in the same terms you do?** Wide
  divergence indicates realignment. Broad agreement plus urgency indicates turnaround.
- **What happened to the last person who raised this?** Repeated raising without
  movement is diagnostic of realignment.
- **Is there slack in the system?** Turnarounds usually have none. Realignments often
  have considerable slack that is invisible because it is fragmented.
- **How is past performance narrated internally?** Pride in a model that the numbers
  no longer support is the signature of a realignment.
- **Who owns the decisions that drive your metrics?** Accountability sitting apart from
  decision rights is structural, and it predicts that operational effort will not move
  the number.

## Where the model is weak

- **Mixed situations are the norm, not the exception**, and the model prefers clean
  categories. Use it to choose a dominant mode, not to force a single label.
- **It says little about functions without line authority.** A leader whose results
  depend on units they do not control faces a coalition problem the model does not
  address directly; assume every situation is harder than its description implies.
- **Situations change underneath you.** A realignment becomes a turnaround the moment
  a result is missed publicly. Re-diagnose at each horizon boundary rather than once.

Block 3 — the-four-quadrants.md

# The Four Quadrants

Figures marked `[estimated]` are judgment, not measurement.

## Why four

A transition plan built as a single list of activities cannot be reasoned about. Work
arriving in a new seat divides into four kinds with different clocks, different risk
profiles and different definitions of success. Separating them stops the urgent from
consuming the important, and makes visible the quadrant a leader is neglecting —
which is usually the first one.

| Quadrant | Clock | Success is | Failure looks like |
|---|---|---|---|
| **Business as usual** | Continuous | Nothing degrades | A miss attributed to the new leader's distraction |
| **Strategic projects** | Inherited | Commitments met or renegotiated deliberately | Discovering an obligation after it slipped |
| **Quick wins** | Quarterly | Visible progress that serves the destination | Wins that later have to be unwound |
| **Transformation** | Multi-year | A credible, sequenced route | A vision with no first step |

## Quadrant 1 — Business as usual

Everything that must keep working while the leader learns. Forecasting, scheduling,
real-time operations, intraday management, adherence, capacity planning — across every
business unit served — plus the standing supplier obligations.

**The test is that nothing broke.** This quadrant produces no credit when it goes well
and consumes everything when it goes badly. A service level miss in the first ninety
days will be attributed to the transition regardless of cause.

What belongs in the plan here is not a description of the work but the **assurance
mechanism**: what is being watched, at what cadence, with what escalation, and who
holds it while the leader's attention is elsewhere. Naming a deputy for BAU is often
the single most valuable line in the quadrant.

Measures usually sit across service level, forecast accuracy, adherence, occupancy,
cost per unit of work, and the experience measures on both the customer and employee
side.

## Quadrant 2 — Strategic projects

Programmes already in flight to which the team is already committed. Migrations,
platform changes, transformation workstreams, integration programmes.

This quadrant is inherited rather than chosen, and it carries a specific hazard: **the
new leader's agenda collides with obligations they did not scope and cannot see.**
Direct reports are frequently committed to programmes whose governance sits outside the
function entirely.

In horizon one the job is inventory and triage, not redirection:

- What is every commitment, to whom, with what date and what named deliverable?
- Which of these did the function actually agree to, and which were assumed on its behalf?
- Where does the function carry accountability without a seat in the governance forum?
- What would happen if a commitment slipped — who finds out, and when?

The last question is the useful one. Commitments whose slippage would be invisible are
where the exposure sits.

Redirection belongs in horizon two or three, once the inventory is complete. Cancelling
a programme in the first thirty days on incomplete information is the most reliable way
to lose the room.

## Quadrant 3 — Quick wins

Near-term improvements that demonstrate progress while the longer work is still being
mapped. Operated as a recurring cadence rather than a one-off exercise — see
`quarterly-okr-cadence.md` for the mechanism.

Two disciplines govern what belongs here.

**Every win must serve the destination.** A visible improvement inconsistent with the
transformation endpoint is worse than no improvement, because it has to be unwound
later and the unwinding costs more credibility than the win earned. The test: *if this
succeeds, does it make the next step easier or harder?*

**Match the win type to the situation.** In a realignment, diagnostic wins count —
establishing a baseline, exposing an unreliable measure, making a hidden cost visible.
But diagnostic wins do not read as wins to a commercial audience, so carry at least one
operational win alongside them, chosen for legibility outside the function rather than
for size.

## Quadrant 4 — Transformation

Where the function is going, and the route. In a workforce and supplier context this is
usually a climb up a maturity model, and it runs on a multi-year clock.

Inside ninety days this quadrant does not move. What it owes is a **credible sequenced
route with an identified first step**, and enough diagnostic work to know the starting
point. A destination without a baseline is a wish.

Two distinct maturity axes apply and should not be collapsed:

- **Delivery maturity** — how well the workforce function performs its own craft.
- **Buyer maturity** — how well the organisation manages its supplier estate. This is
  a different capability, assessed against different standards, and an organisation can
  sit at very different levels on the two.

Most published maturity work addresses the first. The second is where merged estates
are usually weakest, because buying capability accumulates by deal vintage rather than
by design.

## The supplier estate runs through all four

A plan that confines the supplier estate to business as usual and transformation is
incomplete. It appears in every quadrant, and differently in each.

| Quadrant | How the supplier estate appears |
|---|---|
| **BAU** | Service obligations, governance cadence, volume and invoice reconciliation, performance reporting whose provenance is often the supplier itself |
| **Strategic projects** | Suppliers are usually inside the major programmes — frequently without the workforce function named in the governance |
| **Quick wins** | Often the fastest wins available, because commercial and routing levers are discrete and contractual rather than requiring build |
| **Transformation** | Buyer maturity, tier design, and which work is eligible to sit where |

The quick-win row is the one most often missed. Where internal improvement requires
system change and months of effort, a supplier-side change can sometimes be made at the
next governance meeting. That asymmetry is worth exploiting early, when visible progress
is scarce.

## Balancing the quadrants

A rough allocation of a new leader's attention across the first ninety days
`[estimated]`, offered as a starting point rather than a rule:

| Quadrant | Horizon 1 | Horizon 2 | Horizon 3 |
|---|---|---|---|
| BAU | 40% | 30% | 25% |
| Strategic projects | 25% | 25% | 20% |
| Quick wins | 15% | 30% | 30% |
| Transformation | 20% | 15% | 25% |

The failure patterns are predictable. A leader who under-weights BAU in horizon one
gets caught by a miss. One who over-weights it never establishes an agenda. One who
starts in transformation produces a strategy nobody can act on, because the baseline it
rests on does not exist yet.

Block 4 — horizon-design.md

# Horizon Design

Figures marked `[estimated]` are judgment, not measurement.

## What each horizon owes

The three horizons are not equal thirds of the same work. Each has a distinct
deliverable, and confusing them is what produces plans that are busy in month one and
empty in month three.

| Horizon | Owes | Does not owe |
|---|---|---|
| **1 — days 1–30** | A learning agenda, a situational call, an inventory of inherited commitments, assurance that BAU is held | Answers, restructures, announced strategy |
| **2 — days 31–60** | First evidence from the learning agenda, the first committed quarterly objectives, a named first win | A finished roadmap |
| **3 — days 61–90** | The route, the operating cadence installed, the first win landed or explained | Delivered transformation |

## Horizon one is a learning agenda, not an activity list

The most common error is filling horizon one with visible activity because thirty days
of listening feels like thirty days of nothing. It is not, and a well-constructed
learning agenda is the most defensible thing a new leader can produce.

Build it as a set of questions with owners and dates — a deliverable in its own right,
reviewable, and demonstrably closed or not closed at the horizon boundary:

- What is the question, stated precisely enough to be answerable?
- Who holds the answer, and have they been asked?
- By what date, and what happens if it is not answered?
- What decision does this question gate? A question gating nothing can wait.

That last test is the useful one. Learning agendas sprawl because every question is
interesting. Only the ones that gate a decision belong in horizon one.

**Also record what is deliberately not being asked yet.** A leader who is visibly
choosing their questions reads very differently from one who is asking everything.

## Horizon two is where evidence first appears

The transition from listening to asserting happens here, and it should be explicit.
Horizon two converts learning-agenda answers into two things: the first committed set
of quarterly objectives, and the first identified win.

The failure mode is arriving at day sixty with a fuller understanding and nothing
committed. Understanding that produces no commitment is indistinguishable, from
outside, from having spent sixty days learning nothing.

## Horizon three delivers a route, not a destination reached

Nothing structural completes in ninety days. What completes is the **plan**, the
**cadence**, and ideally **one landed win**.

The cadence matters more than it looks. Installing a recurring rhythm — quarterly
objectives, a standing review, a reporting line that produces the same artifact every
period — is what converts a transition plan into an operating model. A leader who
lands a win but installs no cadence has to start from nothing next quarter.

If the first win has not landed by day ninety, say why in one sentence and state the
new date. An explained miss is survivable; a silent one is not.

## Starting off-cycle

Almost all transition guidance assumes a clean start. Most real starts land mid-cycle:
partway through a goal year, mid-quarter, or into a planning round already in progress.
The mismatch is normal and should be handled openly rather than smoothed over.

**Name the mismatch in the narrative.** State the start date, where the horizons fall,
and where the organisation's goal cycle ends relative to them. A plan that quietly
implies alignment will be read against the real calendar by someone who notices.

**Do not restate the annual goals as the ninety-day plan.** They operate on different
clocks and answer different questions. The ninety-day plan should show how its horizons
*serve* the annual goals — which usually means establishing the conditions the annual
goals depend on, since those goals were often written before the leader arrived.

**Anchor horizons to committed milestones, not to round numbers.** If a governance
readout, a contract renewal, a planning submission or a review sits at day seventy-two,
that is the real boundary. Horizons built on thirty-day arithmetic will cut across
commitments that already exist, and the plan will be quietly rescheduled by other
people's calendars.

**Where the goal year ends inside the ninety days**, horizon three is a handover into
the next cycle rather than a finish line. Treat it as the place where the next period's
objectives are set, and say so.

**Prorating.** Where performance is assessed against a partial period, raise it early.
A leader assessed on a full year's goals having held the seat for a third of it is a
conversation better had in week three than at calibration.

## Sequencing constraints worth respecting

Some orderings are not preference but dependency. Getting these wrong produces work
that has to be redone.

- **Definitions before comparison.** Comparing units before measurement definitions are
  agreed invites a challenge to the method, and the challenge will be justified.
- **Baseline before target.** A target set on an unestablished baseline will move when
  the baseline is established, and the movement will look like failure.
- **Inventory before redirection.** Cancelling or redirecting a programme before the
  commitment inventory is complete is the most reliable way to discover an obligation
  the hard way.
- **Redeployment before hiring.** Hiring is a long-lead expression of a standard not yet
  set. Re-scoping existing people against the emerging standard is the near-term lever,
  particularly where a merge has created duplication.

## Reviewing at boundaries

Treat each horizon boundary as a checkpoint with three questions:

1. **Has the situation changed?** Re-run the diagnosis. Realignments become turnarounds
   when a result is missed publicly.
2. **What did the learning agenda close, and what did it open?** Open questions carried
   forward without comment are how plans quietly become fiction.
3. **Which quadrant is being neglected?** It is almost always business as usual, and the
   consequence arrives without warning.

Block 5 — quarterly-okr-cadence.md

# The Quarterly OKR Cadence

Figures marked `[estimated]` are judgment, not measurement.

## What this is for

Quick wins fail as a one-off exercise. Surfaced once, they produce a burst of activity
and then nothing, and the function reverts to whatever it was doing. Operated as a
recurring quarterly cadence, they become the mechanism by which improvement is
continuously generated, selected and evidenced.

The instrument is a quarterly objectives-and-key-results cycle owned by the function's
leadership team. It **sits alongside** the organisation's annual goal process and
midyear review; it does not replace them. Say this explicitly and early, or it will be
read as a competing performance system and resisted by the people who own the existing
one.

| | Annual goals | Quarterly OKRs |
|---|---|---|
| Owner | The organisation's performance process | The function |
| Horizon | A year | A quarter |
| Purpose | Accountability and assessment | Focus and generation |
| Tied to compensation | Usually | **No** |
| Expected attainment | High | Deliberately not certain |

The compensation row is the load-bearing one. OKRs work because they can be ambitious,
and they can be ambitious because missing one is not punished. Attach them to pay and
they become conservative within a single cycle, at which point they are annual goals
with a shorter clock and no additional value.

## Structure

**Objective** — qualitative, directional, time-boxed to the quarter. It answers *what
are we trying to achieve, and why does it matter this quarter?* It should be readable
by someone outside the function.

**Key results** — three to five per objective, numeric, and expressed as outcomes
rather than activity. "Publish the definition set" is activity. "Definition set signed
by all unit leaders" is an outcome. The distinction is whether achievement depends only
on the team's own effort; outcomes usually require someone else to agree, which is the
point.

Every key result needs a baseline. A key result with no starting value cannot be graded
and will be argued about at quarter end.

## The generation funnel

Each leader in the team brings **five to ten candidate ideas** and settles on **three**.

The two-stage structure is deliberate, and collapsing it is the most common way the
cadence degrades. Generation and selection are different cognitive tasks. Asking for
three directly produces the three most obvious and defensible ideas — the safe ones,
which are rarely the valuable ones. Asking for ten produces range, and the selection
step then does its job against criteria rather than against the fear of proposing
something that might fail.

Selection criteria, applied in order:

1. **Does it serve the transformation destination?** A win that makes the next step
   harder is disqualified regardless of size.
2. **Can it be evidenced?** If completion would be a matter of opinion, it is not a key
   result.
3. **Is it within the team's power to attempt?** Dependencies are acceptable; complete
   dependency on another function is not.
4. **Does it map to a value dimension?** See below. An objective mapping to none is a
   task, not an objective.

## Mapping to the value spine

Every objective must name the dimension it serves:

| Dimension | What it covers |
|---|---|
| **Customer experience** | Service outcomes as the customer encounters them |
| **Employee experience** | Conditions, load, and sustainability for the workforce |
| **Expense** | Cost of delivery, per unit and in total |
| **Automation** | Work removed, augmented, or made self-serving |
| **Revenue protection** | Revenue retained as a consequence of the above |

Two cautions.

**Revenue protection is downstream.** It is a consequence of experience and expense
decisions rather than an independent lever, so an objective claiming it should name the
mechanism through which it acts. Otherwise the same benefit is counted twice — once as
experience and again as revenue.

**Employee experience is a leading indicator, not a soft one**, particularly where
teams are being merged. Attrition, adherence and quality all move ahead of the customer
measures. An OKR set with no employee dimension in a merging function is incomplete.

## Aggregating across the team

With a leadership team of five — say four workforce leads and one supplier management
lead — three objectives each produces a catalogue of fifteen.

**The catalogue is a portfolio, not a workload.** Not all of it will run. Its value is
breadth: it shows what the function believes is available, it surfaces the same idea
arriving from three directions, and it makes visible which value dimensions nobody
proposed anything against. That last signal is usually the most informative output of
the whole exercise.

**Where one lead holds a disproportionate share of the organisation**, their three
objectives will dominate the catalogue's weight while occupying the same three slots as
everyone else's. Handle it explicitly — either weight contributions by scope, or allow
that lead more slots and say why. Left unmanaged, the catalogue systematically
under-represents the largest part of the function.

**De-duplicate before selecting.** The same objective proposed by three leads is a
signal about priority, not a redundancy to delete. Merge it and note who proposed it.

## Running the cycle

| Timing | Activity |
|---|---|
| Week 1 of quarter | Individual generation — five to ten candidates each, done alone before any group discussion |
| Week 2 | Session: share, de-duplicate, select three per lead, map to value dimensions, set baselines |
| Weeks 3–11 | Delivery. One standing checkpoint mid-quarter, short, no re-planning |
| Week 12 | Grade, retrospective, carry-forward decisions |

Individual generation **before** group discussion matters. Ideas generated in a room
converge on whoever speaks first, and the range that made the funnel worthwhile is lost.

## Grading

Score each key result from 0 to 1. Aggregate to an objective score.

Consistent scores near 1.0 indicate objectives set too safely. A healthy distribution
averages somewhere around 0.7 `[estimated]`, and a quarter with nothing below 0.5
suggests the ambition dial needs moving. Say this at the outset, because a team scored
below 1.0 for the first time will assume it has failed.

Grade in public, within the team. The retrospective is worth more than the score.

## Expect the first cycle to be poor

The first quarter's objectives will be too many, too vague, and too much like activity
lists. This is normal and not a reason to abandon the cadence. The second cycle is
markedly better because the team has now graded the first one and understands what a
weak key result feels like at quarter end.

Do not attempt to fix the first cycle by rewriting it mid-quarter. Let it run, grade it
honestly, and let the retrospective do the teaching.

Block 6 — boss-conversation.md

# The Conversation With Your Boss

Figures marked `[estimated]` are judgment, not measurement.

## The plan is a conversation, not a document

A transition plan shared upward is one move in a sequence of conversations, and knowing
which one it is changes how it should be written. Michael Watkins identifies five
distinct conversations a new leader needs with their boss, and the failure mode is
attempting all five in one meeting, where they collapse into a status update.

| Conversation | What it establishes | When |
|---|---|---|
| **Situational** | Agreement on what kind of situation this is | Earliest — everything else depends on it |
| **Expectations** | What success looks like, by when, measured how | Once the situation is agreed |
| **Resources** | What is needed to deliver what was just agreed | Immediately after expectations, never before |
| **Style** | How the two of you will work — cadence, channel, escalation | Early, and revisited |
| **Personal development** | Where the role stretches you | Later, once trust exists |

**A ninety-day plan is primarily the expectations conversation, with the situational
conversation embedded in it.** If the situational agreement has not happened, the plan
will be received as a set of assertions rather than as a proposal, because the reader
has not agreed to the premise it rests on.

The resources conversation is the one most often skipped, and skipping it is expensive:
commitments accepted without the resource discussion become unfunded commitments that
are still owed.

## The one-page narrative

The grid is the working artifact. The page is what gets read. Four sections.

**Situation.** Two or three sentences on what has been inherited, framed structurally
rather than personally. In a realignment this section is doing the heaviest lifting in
the whole document — it is where agreement on the premise is either won or lost. Attribute
difficulty to structure and history, never to predecessors.

**What will be proven, by when.** Three horizons, two or three lines each, each with an
observable. Not activity — evidence. "A baseline placement with traceable evidence
behind every score, by 19 September" rather than "complete the assessment."

**What is being asked for.** Decision rights, access, named partners in adjacent
functions, a steer on scope questions, budget where relevant. Make these specific and
few. A plan that asks for nothing signals either that nothing is needed — rarely true —
or that the leader has not worked out what is.

**What is explicitly out of scope.** The highest-value section, and almost always
omitted. Naming what will *not* be attempted in ninety days converts an unbounded plan
into a bounded commitment, and it pre-empts the most common failure conversation, which
is about something the leader never intended to do but never said so.

## What to commit to, and what not to

**Commit to establishing things.** Baselines, definitions, measurement, inventories,
cadences. These are within the leader's control and demonstrably complete or not.

**Do not commit to moving a number that cannot yet be measured defensibly.** A committed
delta on a figure whose provenance is weak commits the leader to the weak version of it.
When the measurement is corrected, the target moves, and the movement reads as failure
regardless of cause.

**Do not restate inherited annual goals as the plan.** They answer a different question
on a different clock. Show how the horizons serve them.

**Do not commit to outcomes gated by other functions** without naming the dependency in
the same sentence.

## Pre-announce the predictable bad news

Some things are certain to look worse before they look better, and they are foreseeable.

Honest normalisation of a measurement usually moves some previously-flattering numbers
in the wrong direction — internal figures are frequently flattered by unmeasured work,
and outsourced figures depressed by the mix of work they receive. Measuring something
properly for the first time routinely produces an apparent regression.

**Say this in writing before the first corrected number lands.** A predicted result and
an unexplained regression are the same event with opposite political consequences, and
the only difference is who described it first. This costs one paragraph and it is the
highest-return paragraph in the document.

## Asking well

The resources conversation goes better when what is being asked for is specific,
justified by the plan rather than by principle, and small in number.

- **Decision rights** where accountability already exists without them. Name the specific
  decisions, not the principle.
- **Named partners** in finance, HR, recruiting and learning. A function wired into
  adjacent functions needs counterparts, and asking for them by role is cheaper than
  discovering their absence in month three.
- **A seat in governance** where the function carries accountability for outcomes decided
  elsewhere. Ask for the forum by name.
- **A steer, not a decision**, on genuinely open scope questions. Offering a default
  position with each question makes it answerable in a sentence and demonstrates the work
  has been done.

## Style, and the cadence question

Agree early: how often, in what form, and what warrants an immediate escalation rather
than waiting for the next review. Agree specifically what the boss wants to hear about
before anyone else tells them — this is the single most useful thing to establish in the
first month, and most new leaders never ask it directly.

Also establish how the boss prefers to receive bad news, and how much finished thinking
they want. Some want the problem early and unformed; others want it with an
accompanying recommendation. Both are workable. Guessing wrong is not.

## Reading the response

The plan's reception carries information worth capturing.

- **Which section drew questions?** The situation section drawing challenge means the
  situational conversation has not concluded, whatever was said.
- **Was anything asked for granted?** A plan whose asks are all deferred is a plan whose
  commitments should be revisited.
- **What was added?** Additions signal expectations that existed but had not been stated,
  which is the most valuable output the conversation can produce.
- **Was the out-of-scope section challenged?** If not, it is agreed — and that agreement
  is worth recording in writing, in a follow-up note, the same day.

Block 7 — the-workstream-spine.md

# The Workstream Spine

Figures marked `[estimated]` are judgment, not measurement.

The four quadrants classify work. They do not organise a document, and using them as its
chapter structure produces a plan that is hard to read and easy to misread. This block sets
out the alternative: **workstreams across horizons as the spine, quadrant as a tag.**

## Why quadrants fail as a spine

Three failures are predictable, and all three are visible to the reader.

**The BAU chapter becomes a dumping ground.** Asked "what is in business as usual?", a new
leader lists everything they are doing that is not a project. The quadrant's actual meaning
— the standing obligation that nothing degrades on the handover — disappears under
activity. BAU is a *commitment*, not a workload. Its entire content is what must not break,
who watches it, and what the leader will be told early enough to act.

**Cross-quadrant work has to be told four times.** A supplier remediation is simultaneously
a BAU obligation, part of an inherited programme, a source of quick wins, and a step on the
multi-year route. Under a quadrant spine it appears four times in four registers, and the
reader cannot tell whether that is one thing or four.

**Sequence disappears.** Quadrants say nothing about order. A plan's argument is that *this*
comes before *that* for a reason — and that argument is what the leader above is actually
reading for.

## The spine

Organise by **workstream**, run each across **three horizons**, and tag each with the
quadrant it serves. Typically five to eight workstreams `[estimated]`; more than that is a
task list, fewer is a slogan.

| Workstream | Horizon 1 | Horizon 2 | Horizon 3 | Quadrants touched |
|---|---|---|---|---|
| ... | learning agenda | first evidence | route + one landed win | BAU · QW |

The quadrant column is what preserves the discipline the quadrants exist to enforce — the
balance check still runs, but across the plan rather than within its chapters. A workstream
touching no quadrant is not a workstream; a plan whose workstreams collectively leave a
quadrant empty has a gap worth naming.

## The action inventory

Separate from the workstreams, and reported as a standing list: the **in-flight actions
that gate discovery**. Site visits being arranged, sessions being calendared, data being
requested, access being granted.

These do not belong inside a horizon. They are not commitments about outcomes and they do
not compete for attention with the workstreams — they are the enabling conditions for
knowing anything at all, and burying them inside a horizon narrative hides both the
dependency and the slippage.

| Field | Why it is there |
|---|---|
| Action | What is being chased |
| Purpose | The decision or discovery it unblocks — an action unblocking nothing does not belong |
| Owner | Named, including where the owner is the leader |
| Status | In progress, blocked, or landed. Blocked is a legitimate status and is worth surfacing |

Two properties make this list valuable to the leader above. It is the part of the plan they
can most easily help with — an escalation, an introduction, a budget approval. And it is
the honest face of a transition: it says what is not yet known and what is being done about
it, which reads as control rather than as gaps.

## Quick wins branch into the roadmap

The most common failure in a transition plan is that the quick wins and the multi-year
destination are unrelated — the wins are chosen for visibility, the destination for
ambition, and no line connects them. A reader who notices this concludes, correctly, that
the wins are theatre.

**Every quick win names the roadmap branch it serves.** Not "consistent with the strategy" —
the specific branch, so that the win is legible as the first increment of something rather
than as an isolated improvement.

This has a second effect that matters more than the first. A multi-year evolution is
difficult to fund and difficult to believe. A multi-year evolution whose first three
increments are already delivering is a different proposition, and the branch linkage is
what converts one into the other. **The wins are the evidence that the route is real.**

Where a candidate win serves no branch, one of two things is true: the roadmap is
incomplete, or the win is opportunistic. Both are worth knowing, and neither is
disqualifying — but the plan should say which.

## Weighting the multi-year evolution

The transformation quadrant carries a presentation risk that is rarely stated: a plan
weighted toward multi-year evolution reads as *"this leader will produce nothing for three
years."* That reading is unfair and it is also predictable, and a plan that invites it has
made an avoidable error.

Three rules `[estimated]`:

1. **Under-weight it in space, not in substance.** The route is real and the plan should
   say so; it does not need more page count than the work delivering inside the ninety days.
2. **Lead with the increment, not the destination.** State the first step and the evidence
   it will produce, then the direction it serves. The reverse order is what triggers the
   misread.
3. **Never present the destination without the near-term branch that reaches toward it.**
   An unattached multi-year vision is the single most reliable way to lose a new leader's
   credibility with the leader above.

The corrective is not to shrink the ambition. It is to make the ambition cash near-term
evidence, which the branch linkage above already provides.

## The OKR cadence as the quick-win engine

Quick wins should not be invented by the leader alone. Running the generation funnel across
the leadership team surfaces what is already on people's plates — which is faster, better
calibrated, and builds ownership that a leader-authored list does not.

Where candidates do not surface, the funnel develops them rather than waiting. The
selection criteria already carry the branch test as their first gate, so a cadence run
properly produces quick wins that are pre-qualified against the roadmap.

The plan should show the *mechanism*, not only its current output. A named cadence that
will keep producing wins is a stronger commitment than a fixed list of three, and it
survives the discovery that any particular win was the wrong one.

## Assembling the document

```
1. Situation and stance          where the organisation is, in one page
2. The workstreams               the spine — each across three horizons
3. Quadrant balance              one table, as a check rather than a chapter
4. Quick wins and their branches the wins, each naming what it reaches toward
5. The route                     multi-year, led by its first increment
6. Action inventory              what is being chased, and what would help
7. What is not yet known         the open questions and what will answer them
```

Sections 6 and 7 are the ones most often omitted and the ones that most reliably earn
trust. A transition plan that states its own gaps is read as an accurate document; one that
does not is read as an optimistic one, and is discounted accordingly.

Usage notes

Sizing. The instruction block is about 470 words and loads with every message inside the project. The six reference blocks total roughly 7,200 words and load only on retrieval.

Packs do not compose at runtime. Each pack is deployed as its own Claude project, so a session running this pack cannot reach another pack's reference blocks. The cross-references in Related packs below are for a human choosing what to deploy next, not instructions the model can follow. Nothing inside Blocks 1–6 refers to material outside this pack, and it should stay that way.

Attribution. The situational model in Block 2 adapts the STARS framework from Michael Watkins, The First 90 Days (Harvard Business Review Press). The five conversations in Block 6 are also his. The adaptation to a combined workforce and supplier function, the four-quadrant structure, the quarterly cadence and the off-cycle handling are not.

Two things the pack deliberately withholds from the artifact. The per-unit situational read stays private: at the level a plan is shared upward, a five-situation matrix reads as complexity and invites an argument about labels applied to other leaders' units. And the pack does not set targets on measures whose provenance has not been established — it commits to establishing them instead.

Article debt. Most of this pack has no canonical article home. The wiki holds three maturity models, which Block 3's transformation quadrant rests on, but nothing on leadership transition, horizon design, quick-win selection, or quarterly OKR practice — a search for each returns nothing. Until those articles exist, these blocks are the canonical version of that material, which is the reverse of the intended dependency.

Related packs

Deploy each in its own project when the work reaches it. They do not combine into one project.

Pack	Deploy it when the plan reaches
Planning Under Demand Volatility (CP-WFM-001)	Business-as-usual forecasting and capacity work under unstable demand
Agent Capability Ontology (CP-WFM-002)	The transformation quadrant's supply-side capability model
Service Quality Comparability (CP-WFM-003)	Comparing quality across internal, captive and outsourced delivery
Content Delivery Under Capacity Constraint (CP-WFM-004)	Getting updates, training or communications to the frontline without buying it out of service level

Change history

Version	Date	Change
1.0	2026-08-11	Initial publication. One instruction block, five reference blocks.
1.1	2026-08-12	Added the-workstream-spine.md as Block 7. Quadrants are demoted from the document's chapter structure to a classification tag; the plan is organised by workstream across horizons. Adds the action inventory for in-flight items that gate discovery, the rule that every quick win names the roadmap branch it serves, and the weighting rules that stop a multi-year route reading as three years before value.

See also

• WFM Maturity Model and WFM Labs Maturity Model — the destination the transformation quadrant climbs
• Vendor Operating Model Maturity — buyer capability, the second and less-examined maturity axis
• Wiki:Packs — the full pack inventory