Build the Rule, Not the Destination

A plan is a list of moves approved once and ageing from that day. A rule is re-run whenever work, supply, objectives, or constraints move, and keeps producing a destination.

Build the rule, not the destination is the design principle that a workforce strategy should consist of a rule that keeps producing a destination, rather than of the destination itself. A strategy that is a list of moves — this many people, in these locations, by this date — starts aging on the day it is approved, because every condition it was built for moves afterward. The principle is distinct from its neighbors. Level 4: Planning in Distributions describes what the plan is at the maturity level where it becomes a distribution. Capacity Planning Cycle describes the monthly mechanics by which an outlook is refreshed and cleared, and Placement Engine Architecture the machinery that produces placements from a constraint register. Strategic Workforce Planning covers the long-horizon activity itself — the roadmaps, budgets and site strategy a rule produces — and Scenario Planning and Contingency Staffing the practice of holding several future states at once. This page concerns the form a strategy takes so that those activities have something durable to run.

Plans and planning

A plan is a set of decisions taken in advance. Planning is the standing capability to re-derive those decisions when conditions move. An operation needs both, and the two are routinely confused, because a plan is a visible document and planning is a capability with no document of its own.

Mintzberg's analysis of strategic planning drew the distinction in its sharpest form. Formal planning, in his account, is the articulation and elaboration of strategies that already exist; it cannot create strategy, because analysis does not produce synthesis, and a process that formalizes a strategy also freezes it against the conditions that will change it. The planners' useful role, in the same account, is to supply the analysis and the catalysis that keep a strategy responsive, not to produce a document that stands in for it.

The pattern is recognizable in workforce strategy. A multi-year sourcing program specifies a destination to the individual site and headcount, due in a stated quarter. The analysis behind it is careful. Then volume, wage inflation, technology and client mix move, and the destination is wrong with no mechanism to re-derive it. The people who built it move on, and nobody closes it out, because that would require re-running an analysis never built to run twice. The program is remembered as a plan that failed; what failed was the absence of planning around it.

What moves underneath a strategy

Four conditions move under any service operation, on different clocks and for different reasons.

What moves	How it moves	What a destination-strategy does when it moves
Leadership	Sponsors change; a program loses the person who could explain why it was shaped as it was	Continues on inertia or stops; nobody owns the re-derivation
Business dynamics	Volume steps up or down; client mix shifts; wage differentials between locations narrow or widen	The cost case the destination rested on is silently invalidated
Service priorities	What the operation is asked to optimize for changes, or the weighting between objectives changes	The destination optimized the wrong thing and cannot say by how much
The human share of work	The proportion of volume a person must touch falls as automation absorbs the easiest work, and the residual hardens	The headcount and skill mix the destination specified no longer match the work

Anything built for one state of these four dies with that state, and a destination is by construction built for one. The only question is how long the state lasts, and in service operations it is rarely as long as the program built for it.

What does not change

Three quantities hold still while everything above moves: revenue, expense, and the profit they net to. They are the terminal objectives — the ones a multi-year strategy can be built against without being rewritten when conditions move. They are not the objective vector a solver optimizes inside a cycle. Level 4: Planning in Distributions specifies that vector — service stability, total cost, agent experience and resilience, selected from a Pareto set — and those are intermediate objectives: stable for a planning cycle, derived from the terminal three, and re-weighted when the terminal three move. That page also notes that a historical service target is not exempt from validation, which is agreement rather than contradiction: the target value is a parameter, while service stability remains an intermediate objective. Everything else that appears in a strategy — the number of sites, the mix of locations, the share of work with suppliers, the share automated, the headcount by skill — is neither. Each is a parameter of the rule or an output of it, and a strategy that promotes one of them to an objective has fixed a variable that should have been free.

Experience is the apparent exception, and it is handled by the service-profit chain. Heskett and colleagues' model links employee satisfaction, service value, customer satisfaction and loyalty to revenue growth and profitability as a causal chain. Where the chain holds, experience is not a fourth terminal objective competing with the three financial ones. It is the mechanism through which they are produced, and it enters the rule as a modeled mechanism rather than as a constraint that is merely respected. The reduction carries a condition that Instrumenting the Objective Before Building the Model makes explicit. The mechanism must carry a measured attribute on every unit of supply; otherwise the rule collapses to the cost-only answer that page describes rather than simplifying to a financial one.

Meadows's ordering of leverage points gives the same result from a different direction. Parameters — the numbers in a system — are the lowest leverage points, and most attention goes to them. The rules of the system and the goals it pursues are far higher, because they determine what every parameter will be set to. A destination-strategy is a set of parameters. A rule-strategy is a rule and a goal, and it produces the parameters.

The rule that keeps producing a destination

A rule, in this sense, has five parts. A stated objective set, held by leadership. A constraint register, with each constraint carrying a strictness class and priced where its class allows — specified at Placement Engine Architecture and populated by the procedure in Translation, Not Negation. A record of the measured attributes of every unit of supply, so that the objectives have something to discriminate on — the supply cards of the engine page, under the input condition Instrumenting the Objective Before Building the Model states. A recompute cadence, on which the rule is re-run — the subject of Capacity Planning Cycle. And a log, so that each destination the rule produces is stamped with the inputs that produced it.

Given those five parts, the rule produces a destination: where work should sit, how many people of which skill, in which locations, under which supplier share, this cycle. The destination is a report rather than a commitment. When work, supply or objectives move, the rule is re-run and produces a different report, and the difference is explained by the inputs that changed. The commitment is to the rule and to its cadence. Nothing in the strategy is due in a stated quarter except the next re-run.

	Destination-strategy	Rule-strategy
What is approved	A set of moves with dates	An objective set, a constraint register and a cadence
What ages	The whole document, from the day of approval	Individual register rows, each re-tested on its own clock
What happens on change	The document is wrong; a new program is commissioned	The rule is re-run; the new destination is explained by the changed input
Who owns it	The program that built it, until the program ends	The function that runs the cadence, indefinitely
What success looks like	Arrival at the destination	A narrowing gap between each cycle's destination and the last, as the inputs stabilize
What survives a leadership change	A deck	A register, a cadence and a log

What the rule does not remove

The principle is not an argument against plans. A hiring pipeline, a supplier relationship, a technology roadmap and a site cannot be improvised when conditions move, and a rule that produced a destination with no capacity to execute it would be a report and nothing more. The principle is about what the plan is derived from. A plan derived from a rule can be re-derived when its conditions move; a plan that stands alone cannot. Mintzberg's planners — analysts and catalysts around a strategy rather than authors of a document — are the staffing model the rule implies.

Nor does the rule settle the objective set. Which objectives are stated, and how they are weighted against one another, is a leadership decision that the rule consumes rather than produces. A rule with an unstated objective set produces the destination that its measured attributes permit, which — as Instrumenting the Objective Before Building the Model shows — is usually a cost-only answer, because cost was the only thing measured everywhere.

Failure modes

Failure mode	What it looks like	Countermeasure
Destination approved as strategy	A multi-year headcount-by-location table is the strategy document	Approve the objective set, the register and the cadence; the table is this cycle's output
Parameter promoted to objective	"Three regional hubs" or "forty percent automated" becomes a goal	Parameters are outputs; the strategy is built against the terminal objectives, with experience as a modeled mechanism
Rule without cadence	The model exists but is run when someone asks	A fixed recompute cadence, owned by a standing function
Rule without log	Each destination replaces the last with no record of what changed	Every output stamped with its inputs; the difference between cycles is reported
Rule without execution	Destinations are produced and nothing is built to reach them	The plan remains; it is derived from the rule and re-derived with it

Maturity Model Position

At Level 2 a monthly cycle already refreshes the outlook, but strategy above it is a destination — a document of sites, headcounts and dates — and it ages as described. At Level 3 parts of the rule exist: a constraint register is kept and checked ad hoc on contested proposals, but the objective set is implicit and the checks are binary, so the strategy is still re-commissioned rather than re-run. At Level 4 the rule is complete and the destination is a distribution rather than a point, recomputed each cycle against the cone of uncertainty. At Level 5 the recompute is automated on defined trigger classes, and a change on any of them produces a new destination without waiting for a cycle boundary.

See Also

• Level 4: Planning in Distributions — what the destination looks like once the rule produces a distribution rather than a point
• Capacity Planning Cycle — the monthly cadence on which the rule is re-run
• Placement Engine Architecture — the machinery that turns the register into a placement
• Translation, Not Negation — how positions become the constraints and weights the rule consumes
• Instrumenting the Objective Before Building the Model — why an unmeasured objective cannot discriminate
• The Flaw of Averages — why a destination built on a single average is wrong on average
• Strategic Workforce Planning — the long-horizon end of the planning spectrum the rule serves
• Scenario Planning and Contingency Staffing — planning for several states rather than one