Holding Company Structure — Decide What the Centre Is For Before You Draw It

The most expensive decision in a holding group is rarely made explicitly. Somewhere between the third and the fifth acquisition, the head office starts doing things for the subsidiaries: buying for them, hiring for them, approving for them. Each step looks efficient on its own. A few years later the operating companies carry the P&L while the levers that move it sit in a building they don't run, and the group is arguing about centralisation one function at a time without ever having answered the prior question. What is the centre for?

We are asked the shared-services question more than almost any other by groups in Saudi Arabia and across the Gulf, many of them family-owned and diversified across trading, real estate, manufacturing and a newer venture or two. The question usually arrives as a list: should HR be central, should procurement, should IT. The list is the wrong starting point, and the rest of this piece explains why.

"I have the P&L but not the levers"

A group we worked with in Saudi Arabia had centralised almost everything into a group services company after a period of fast acquisition: procurement, HR, IT, legal and finance operations. The logic was sound on paper. The group was paying for five finance teams, five IT contracts and five different ways of approving a purchase order.

Two years later the food-manufacturing subsidiary was losing production days waiting for raw materials, because group procurement ran every purchase through an approval process designed for real-estate contracts. The technology venture could not hire engineers, because group HR applied a pay structure built for trading-company roles. The manufacturing CEO put it better than we could: "I have the P&L but not the levers. When I miss the number, I'm accountable for decisions I didn't make."

The group did not have too much centralisation or too little. It had centralised by function when it should have centralised by activity, and it had never decided what role the parent was meant to play.

The Three Roles of the Centre

A holding company's head office can play three roles, and each one adds a layer on top of the one before it. Deciding which of them your centre plays is the design decision; the org chart follows from it.

• Owner. The centre allocates capital between the businesses, manages treasury and group funding, runs audit and risk, and appoints and holds accountable the people who lead each subsidiary. Every holding company plays this role, because it is the reason the group exists.
• Architect. On top of ownership, the centre sets standards the businesses share: group brand, governance and delegation policies, the talent pipeline for senior roles, a common way of reporting performance upward. It designs the rules and leaves the running to the subsidiaries.
• Operator. On top of both, the centre runs activities on the subsidiaries' behalf as shared services: payroll processing, the general ledger and consolidation, IT infrastructure, routine legal and company-secretarial work. This is the role with the largest potential savings and the largest potential for the frustration described above.

The three roles map onto activities far better than onto functions. "HR" is not one thing: processing payroll centralises well, while recruiting specialist engineers for a technology venture almost never does. "Finance" splits the same way, between consolidation (central) and commercial pricing decisions (local).

The Buy-It-Back Test

Once a group has chosen its roles, every candidate for a shared service should pass a single question, asked from the subsidiary's side of the table: if this subsidiary could buy the service from an outside provider at the same cost, would it still choose the centre?

If the answer is yes, because the centre knows the business, holds the data, or genuinely delivers faster and cheaper, the shared service is earning its place. If the answer is no, the service is a tax. It may still be necessary for control reasons, and that is a legitimate choice, but then it should be named as a control and governed as one rather than sold to the subsidiaries as efficiency.

Here's the contrarian part

The usual debate is between centralisation, which promises savings, and decentralisation, which promises speed. We think that framing is why so many groups oscillate, centralising in one strategic cycle and handing everything back in the next. The real mistake is centralising whole functions instead of specific activities. A group that moves "procurement" to the centre bundles indirect spend, where scale genuinely pays, with direct materials for a factory, where a one-week delay costs more than any discount saves.

The second contrarian point concerns uniformity. Groups often insist that every subsidiary adopt the same org chart and the same job titles, on the argument that it simplifies governance. It simplifies the head office's slides, which is not the same thing. A contracting business and a software venture should not share a structure. What they should share is the governance layer on top: how authority is delegated, which appointments the parent controls, and how performance is reported. That layer is where consistency pays, and it is also where most groups have the least of it written down, which is how decision drift sets in across a portfolio.

Why it works

Goold and Campbell's work on corporate strategy is the clearest foundation for this. In Strategies and Styles they described three ways parents manage their businesses, broadly financial control, strategic control and strategic planning, and showed that each could work when it fitted the portfolio. Their later work with Alexander added the harder standard: a parent must add more value to its businesses than it costs them, and more than a different owner could add. They called it parenting advantage, and it is the logic behind the Buy-It-Back Test.

The same logic explains why portfolio shape matters. A group whose businesses are similar can afford an operator centre, because one set of shared services really does fit all of them. A group spanning very different industries usually cannot, and is better served by a strong owner and architect with a thin operating layer. If you are still deciding which businesses belong in the portfolio at all, the BCG matrix is the tool for that conversation, and it should come before this one.

Finally, none of this holds without written governance. The distinction between an owner who sets the rules and an operator who executes them has to be documented as delegated authority, reserved matters and committee mandates, or it collapses back into whoever phones the group CEO first. That is the ground our piece on corporate governance covers, and it is also why chart and structure are not the same thing, as we explain in org chart vs organizational structure.

A practical checklist

• Write one sentence stating which of the three roles your centre plays. If the executive team cannot agree on the sentence, stop there; the org chart will not settle it.
• List activities, not functions. Break each function into the activities it performs and decide each one separately.
• Run the Buy-It-Back Test on every shared service with the subsidiary CEOs in the room, not only the service heads.
• Publish service levels and the full cost of each shared service, and review them with subsidiaries at least annually.
• Separate controls from services. Where the centre keeps an activity for control reasons, say so and govern it as a control.
• Standardise the governance layer, not the org charts: delegation of authority, reserved matters, the parent's appointment rights, and upward reporting.
• Revisit the design after each significant acquisition or disposal. The right centre for three similar businesses is rarely right for six different ones.

Ask yourself

• If a subsidiary CEO misses their number, how many of the decisions behind it did they actually control?
• Which of our shared services would a subsidiary keep if it were free to buy elsewhere at the same price?
• Have we ever written down which of the three roles our head office plays, and would each subsidiary CEO describe it the same way?
• Where have we centralised a whole function when only one activity inside it gained from scale?
• Which approvals travel to the group because of a documented rule, and which travel there out of habit?

The takeaway

Decide what the centre is for before you decide what sits in it. Choose whether the parent plays owner, architect or operator; centralise specific activities rather than whole functions; and keep only the shared services a subsidiary would buy back if it had the choice. Everything else belongs with the people who carry the P&L.