ROC Organization Models

The two-layer hybrid: one planning partner per business leader over one team per planning horizon. Leaders see one face; departments see one counterpart per question. Two layers on two different axes is what makes it a hybrid.

ROC organization models are the shapes a workforce management function can take once it is understood as a resource optimization center — the coordinating node through which work placement flows across an enterprise — rather than as a department that produces schedules. Four shapes recur: functional, organized by the stages of the planning cycle; business-solution, organized by segment, brand or product; regional, organized by geography; and hybrid, which combines two axes in layers. The choice is not a preference. It decides how many counterparts each department in the interface register must deal with, how much capacity can be pooled across the estate, and whether a consolidated function converges on one method or ratifies several. This page sets out the four models, the two tests that decide between them, and the hybrid most consolidated estates land on.

Two objects that are routinely confused

Before the models, a distinction that dissolves much of the debate. How demand is segmented for planning and how people are organized are different objects. A servicing-treatment taxonomy — work that is deflected, work that is contained by automation, work that reaches a person — is a demand model that any organization must plan against; it is not a reporting line. Treating a planning taxonomy as an org chart produces a structure that must be redrawn every time the taxonomy's boundaries move, which for automation tiers is every release. The models below are about people. Whichever wins must plan against the same demand taxonomy.

The four models

Model	Organized by	What it gives	What it costs	Interfaces
Functional	The stages of the planning cycle: long-range capacity → forecasting → scheduling → intraday and real-time → analytics, plus placement	One methodology and one source of truth; specialization by stage; one owner per platform; pooling across the whole estate	No single face for a business leader; segment nuance can be flattened; hand-offs between stages inside the function; risk of a planning factory detached from the operation	Each department has one counterpart for the kind of question it asks. Business leaders have several
Business-solution	Segment, brand, product or client tier, each with its own planning stack	One accountable planning leader per business; room for segments whose planning physics genuinely differ (illustrations: a named-book premium service, a segment that staffs to a containment rate, a public-sector client on a statutory fiscal calendar)	The stack is replicated per segment; methods can diverge; pooling across segments is limited	Business leaders have one face. Each department has one per segment and reconciles across them
Regional	Geography: continents, countries, delivery hubs	Time zone, language and employment law are handled where they arise; follow-the-sun coverage has regional presence	Methods can diverge by region; pooling across regions is limited; demand that is global in origin is planned in regional pieces	Regional leaders have one face. Global departments (finance, technology, commercial) have one per region
Hybrid	Two axes in layers: one for the interface, one for the work	One face per leader and one method for the estate, when the layers are on different axes	Carries a matrix; the layer boundary must be designed, not left to emerge	Designed explicitly — the point of the model

The faces arithmetic

The mechanism by which interconnection decides organization is countable. Take the nine interfaces of the register and the number of planning units the function is divided into. Under a functional model each department has one counterpart for each kind of question — finance deals with the capacity seat, recruiting with the forecasting seat — so the department-side count is one per interface, while each business leader deals with as many planners as there are stages. Under a business-solution or regional model with n units, each department has n counterparts for the same question and must reconcile n answers, while each leader has one. A function with four segments therefore gives finance four headcount plans to reconcile at every close and recruiting four requisition streams to sequence, which is the concrete form of the reciprocal-interdependence cost the register describes. The hybrid exists to make both counts one: one face per leader in the thin layer, one counterpart per question in the execution layer. Any organization-model decision can be tested by writing the two counts down.

Which model, when

The judgment this page makes, stated as a judgment. The functional model is the only one of the four that produces a single method and a single source of truth, and it is therefore the one that lets a merged function converge rather than ratify its heritages; its weakness is the leader interface, which the hybrid repairs. The business-solution model is the right answer only for segments with a permanent claim to different planning physics, and in a merged estate it is usually the inherited state, so choosing it deliberately is choosing not to integrate. The regional model is warranted at the execution layer where the pooling test says work is place-bound, and rarely at the planning layer. The hybrid is where consolidated estates that run both tests tend to land.

The two tests that decide

The pooling test. How much of the estate's work can be served from anywhere, and how much is bound to a place by language, regulation, data residency or an on-soil promise? The ratio of place-bound to poolable work decides how regional the organization must be, and it is best obtained by segment and by country, at which point the regional debate becomes arithmetic. Place-binding is the eligibility constraint the sourcing pages treat as a flag that shrinks the set of admissible locations (Sourcing Design Axes: Node and Client Ownership). Two qualifications matter. The ratio should be treated as a curve rather than a number: the working assumption, to be tested estate by estate, is that regulatory and data-residency constraints do not decay while language constraints may as translation and assistance tools mature — so the organization is designed for the constraint set expected at the end of the planning horizon, not today's. And where the ratio says regional presence is needed, it is usually needed at the execution layer, not the planning layer.

The size test. How large does the human organization need to be at all? The interior optimum — the containment level beyond which further automation costs more in escalation and complexity than it saves — sizes the human population, and that number is upstream of every structural question. A function designed for a population that automation is about to halve is designed for the wrong estate (The Hardening Residual).

The hybrid of record: two layers on different axes

Consolidated estates that run the two tests tend to land on the same shape, described in operating detail at The Workforce Broker: a thin leader-facing layer of planning partners aligned to business leaders, over an execution layer organized by planning horizon and platform.

• The leader-facing layer is senior, small and cheap to re-point. It gives each business leader one counterpart, carries the leader's outlook into the clearing cycle, and returns the scored proposal. The broker page's rule is that this layer is indexed on the portfolio map — which is volatile, so partners are re-pointed at each reorganization — while the internal structure is indexed on stable properties of the work.
• The execution layer holds the stack once, organized by planning horizon — long-range, forecasting, scheduling, intraday and automation, placement — each seat owning its platform. The core role families are at WFM Roles; the horizon organization and the placement seat are described at The Workforce Broker. This layer is where pooling happens and where the single method lives.

The matrix this creates is deliberate and is held inside the function. The alternative — each leader facing several planners and each planner facing several leaders — is the same matrix held by everyone else. Where an estate is mid-migration, the design rule that keeps the matrix stable is that planning follows the account while support follows the platform: an account's outlook moves to the receiving partner the day it is reassigned, and the execution team keeps running it on the legacy platform until the migration's own schedule moves it.

Where the function reports

Independent of its internal shape, the function reports somewhere, and the real-time ROC page describes three placements with their costs: embedded in workforce management (tight to the planning cycle, limited authority to direct change across departments), embedded in operations (broader cross-functional authority, distance from the analytical rigour of the planning function), or independent (equal standing with every function, at the cost of governing the duplication it creates). Read against the interface register, this page adds one observation: embedding in operations also lengthens the finance and technology interfaces, and independence serves every interface equally only if the function is given authority at commitment points. The integrated center resolves it by centralizing the picture and the machinery while leaving decision rights with the owners of outcomes — which is a statement about authority, not about reporting line.

Failure modes

Failure	Mechanism	Test that catches it
Residue mistaken for design	A merged function inherits one heritage's vertical stack beside another's horizontal one; the collision is never decided, so the migration decides it	Name each leader's organizing axis; where there are three, there is no design yet
Regionalising around a decaying constraint	Regional units built for a language constraint that assistance tools are dissolving	The pooling test run on the end-state constraint set
Taxonomy as org chart	Demand tiers become reporting lines and are redrawn every release	Separate the demand model from the people model
Segment structure as integration avoidance	Business-solution alignment chosen because it is the current state	Ask which segments have a permanent claim to separation; usually one
Two layers on one axis	Partners and execution both aligned to segment, so the hybrid is a functional model with extra hand-offs	The layers must differ in axis or there is no hybrid

Relation to organization design theory

The models are the workforce-function form of the classical structural choices — functional, divisional by product or market, geographic, and matrix — and the hybrid is the matrix form with one axis deliberately kept thin. The underlying choice is the one organization theory frames as grouping by function versus grouping by market — product, client or place — with each basis buying coordination within the group at the cost of coordination across it. Two of the design tests from the strategy literature apply directly: whether units that need a distinct specialist culture are protected from a dominant one, and whether the difficult links between units that must cooperate are designed rather than assumed. The interface register is how the second test is passed in this domain. The systems view adds the reason the choice matters more here than in most functions: the loops that run through recruiting, training and attrition carry delays of months, and a structure that splits those interfaces across several units adds decision delay to pipeline delay — the condition under which a controlled system overshoots and undershoots its target rather than settling.

Maturity Model Position

At Level 2 the function is usually organized by whatever it inherited. The functional model is the Level 3 form, because a single methodology and clean platform ownership are what the automation layer needs. The two-layer hybrid is the Level 4 form, when placement genuinely moves and business leaders need one counterpart for a scored proposal. At Level 5 the execution layer is largely machinery and the leader-facing layer is the organization.

See Also

• Interconnected Workforce Management — the interface register the models are tested against
• The Workforce Broker — the two-layer design in operating detail
• Integrated Global Resource Optimization Center — the integrated center and what it centralizes
• Resource Optimization Center (ROC) — the real-time center and its three reporting placements
• WFM Center of Excellence CoE Design — the center-of-excellence form
• WFM Roles — team structure by scale and the core role families
• Future WFM Operating Standard — the GRPI-T pillars as practice
• Systems Thinking — why fragmented interfaces oscillate
• GRPI-T Framework — the frame this design work sits in