Shared Services versus Embedded Teams

What this topic is

Every specialist capability in a company sits somewhere on one axis: fully embedded in each business unit, fully centralised into one shared function, or one of several hybrids in between. The choice is not about org-chart tidiness. It determines whose priorities the specialist serves, how much context they carry, how efficiently their time is used, and who is responsible for making them better at their job. Companies swing between the two poles on roughly a four-year cycle, usually because the last swing solved one set of problems and created the other.

Why it matters

The wrong placement is expensive in ways that never appear as a line item. Embedded specialists in small units are underused, isolated, and slowly deskilled. Centralised specialists are efficiently used, well-supported, and progressively less useful, because the context that made their work good stayed in the business unit. Both failures take about two years to become obvious, by which point the reorganisation that caused them is remembered as a success.

Who should learn it

Executives who own a function that serves multiple business units; general managers arguing to keep or gain their own specialists; and functional leaders trying to build craft standards across people who do not report to them.

What you will understand

• The pooling arithmetic: why five half-time specialists cost more than three full-time ones
• Context cost, the offsetting force, and how to estimate the crossover
• Why the reported savings from centralisation are usually overstated, and by roughly what
• The three things a hybrid must specify or it becomes the worst of both

Prerequisites

• Empire Building and the Internal Capital Market Distortion
• Managing Managers and the Second-line Problem

Common misconception

"Centralising the function will save money through scale, so the question is just whether the business units will tolerate it." The pooling saving is real but it is smaller than the business case says, because a share of the specialists reappear inside the business units within eighteen months under a different job title — hired to do the coordination, translation and pre-work that the central team now requires before it will start. The honest comparison is not central headcount versus the old embedded headcount; it is central headcount plus the shadow staff plus the requesting overhead versus the old number. Companies that measure this find the saving is real and roughly half of what was promised, and the ones that never measure it reorganise again in four years for the opposite reason.