Divisional Structure

What a divisional structure means

A divisional structure (الهيكل التنظيمي بالوحدات) divides an organisation into divisions that stand on their own, each of which gathers the functions it needs to run a whole product, market or region from start to finish.

It stands opposite the functional structure, in which similar functions are grouped into one department that serves the whole organisation. The difference between the two lies not in job titles but in where decisions sit and who answers for the result. In a divisional structure the head of a division holds the functions the division needs, apart from any kept at the centre, and answers for its output. In a functional structure the head of a department holds a specialism and answers for its quality.

The axis a divisional structure is built on

A divisional structure needs a single axis to be built on. Three axes are set out below:

• By product. One division for each product line. This suits a business whose products differ so much that a person who has mastered one has not mastered the other.
• By customer or market. One division for government customers, another for companies and a third for individuals. This suits a business where the sales and service cycle differs clearly between these groups.
• By region. One division for each territory. This suits a business where nearness to the market is a condition of the work, and where regulations or trading customs differ between regions.

The test that favours one axis over another is a single question: along which axis does what the customer needs differ more? If a customer in Riyadh and a customer in Jeddah ask for the same thing, while the buyer of the first product asks for something different from the buyer of the second, the axis is product, not region.

The sign of a wrong axis is two divisions meeting at the same customer with the same offer. In that case the work has not been divided. A team that was working on one thing has been split in two, and the friction that follows is not a failure of anyone’s management.

A divisional structure does not have to move every function at once

Not every support function has to move into the divisions. The rule for ordering them is arithmetic: for each function, compare how many people one division would need with how many would be enough for the whole organisation centrally. The difference is the price of moving that function.

The examples that follow assume an organisation of four divisions and a loaded cost of SAR 15,000 per person per month, or SAR 180,000 a year. These figures, like every figure in the examples, are assumed in order to show the mechanisms. They are not a benchmark and are not drawn from any source.

• A function that needs 3 people in each division and 7 centrally: 12 people across the four divisions, a duplication of 5 people, costing SAR 900,000 a year.
• A function that needs 2 in each division and 6 centrally: 8 people, a duplication of 2, costing SAR 360,000 a year.
• A function that needs 1 in each division and 3 centrally: 4 people, a duplication of 1, costing SAR 180,000 a year.

The third function costs a fifth as much to move as the first (SAR 180,000 against SAR 900,000). A practical order follows from this. Functions with a small minimum per division move to the divisions first, while functions with a large minimum, where pooled expertise weighs more heavily, stay central.

The result is a mixed form: divisions hold what touches their daily work, and the centre keeps what calls for scarce expertise or for a single standard. The choice is not made once. It is revisited whenever the size of the divisions changes, because the three duplication costs above move with it.

The first cost of a divisional structure: the same function, repeated

Because each division gathers what it needs, what was one becomes four in an organisation with four divisions. Take the first function above in detail. A central team of 7 people is enough for the whole organisation, and each division would need 3 if it worked alone:

• The total under a divisional structure: 3 multiplied by 4, or 12 people.
• The excess over the central team: 5 people.
• At a loaded cost of SAR 15,000 per person per month, the excess is SAR 75,000 a month and SAR 900,000 a year.

That is for one support function. If the same figures repeat in three support functions, the total reaches three times that amount. The repetition follows from the nature of this structure, not from a mistake in applying it. How the loaded cost per person is built up is a separate calculation.

What a division has to gain to recover the cost of a divisional structure

The SAR 900,000 excess, spread across four divisions, comes to SAR 225,000 for each. If a division’s revenue is SAR 12,000,000, its independence has to produce an improvement of 1.88 percent of that revenue (225,000 divided by 12,000,000 is 1.875 percent, rounded to two decimal places).

This is the form in which the argument can be settled. The question is not whether a divisional structure is better. It is whether bringing decisions closer to the market in this particular division can reasonably produce an effect of close to two percent of its revenue. Someone who knows the market can judge that question.

The general question cannot be judged in the same way. Whether a divisional structure is better than a functional one is not settled by the definition, and the answer depends on how far the markets differ, how quickly they change and how large the divisions are.

The minimum size at which a divisional structure holds

The support team inside a division has a floor it cannot go below however small the division becomes, so its share of the division’s headcount grows as the division shrinks. With the floor of 3 people from the first function:

• A division of 120 employees: the 3 support staff are 2.5 percent of its headcount.
• A division of 60: 5 percent.
• A division of 20: 15 percent.

Dividing into small divisions therefore spends roughly a seventh of a division’s headcount on support, a share that a division of that size may struggle to carry. That is the sense in which a divisional structure has a minimum size. Below it, a division becomes a name attached to a manager and a few staff while support stays central in practice, so the organisation pays the cost of the form without its benefit.

The calculation shows how the minimum size of a division is derived, but it does not fix a number for that size, because the minimum support team differs from one line of business to another. One indicator that exposes the problem is a very narrow span of control among the managers of small divisions.

Under a divisional structure, the allocation rule decides which division is profitable

Each division answers for its result. Where that result is profit, the division is measured as a profit center is, and the figure holds only after the division has been charged its share of shared expenses. The rule for charging them is not an accounting detail, because it decides the verdict.

Take a shared expense of SAR 2,400,000 a year and four divisions, with 100 employees and SAR 40 million of revenue between them:

• The first division: 40 employees and revenue of SAR 12 million.
• The second division: 20 employees and revenue of SAR 18 million.
• The third division: 30 employees and revenue of SAR 6 million.
• The fourth division: 10 employees and revenue of SAR 4 million.

Allocated by headcount, the second division’s share is 20 of 100, or SAR 480,000. Allocated by revenue, its share is 18 of 40 million, or SAR 1,080,000.

The difference is SAR 600,000 on a single division. If its profit before allocation were SAR 900,000, it would show a profit of SAR 420,000 under the first rule and a loss of SAR 180,000 under the second.

The same division in the same year shows a profit or a loss depending on one line in the allocation policy. For that reason the rule is decided and announced before the year begins, not after the results appear, and it is kept for several years so that a division is compared with itself rather than with a rule that has changed. Choosing the basis is a decision in cost allocation that each organisation takes for itself, and the definition of a divisional structure does not choose it.

What a divisional structure gains in return

• A shorter distance to a decision. What used to pass through three parties passes through one, and the gain can be felt in markets that change quickly.
• Clear responsibility. A division’s result is attributed to one person, which leaves less room for argument over who answers when the result goes wrong.
• General manager development. A division head runs several functions, and that experience cannot be gained inside a specialised department. Where it fits on a career path is a question for job family design.
• Separability. A division that stands on its own is easier to close, sell or add, because its boundaries are known.

What undermines a divisional structure in practice

• Divisions without authority. The structure is announced while decisions stay at the centre, so the organisation pays for the duplication and decisions remain as slow as before. Which rights pass to the divisions is a separate question, the one decentralized management covers.
• Divisions competing for the same customer. Two divisions selling to one customer with no rule to separate them produce what the customer sees as a conflict, and this can surface early in a divisional structure.
• Expertise that stops moving. Specialists spread across four divisions do not meet, so what one division learns may not reach the others, and practices drift apart until moving an employee between divisions becomes difficult.
• Measurement over a short horizon. A division head is held to account for the year’s result and so may defer whatever will not pay off within that year. This effect on incentives follows from the structure, not from the people in it.

Divisional structure beside matrix organization, the organisation manual and the headcount plan

Each of these answers a different question about how an organisation is arranged, and setting the questions side by side is what separates them.

• Matrix organization. A matrix combines the two axes, so an employee reports to two lines. A divisional structure separates them: an employee belongs to one division and reports to that division alone. The matrix resolves the duplication problem and creates the problem of dual reporting, and the divisional structure does the reverse.
• Organisation manual. A manual is a document describing the organisation as it has settled, whereas a divisional structure is a choice about the organisation itself. The document is written after the decision, and it neither replaces the decision nor creates it.
• Headcount plan. A headcount plan decides how many employees there are and where they sit, whereas a divisional structure decides the basis on which positions are grouped. The plan is built after the form of the structure is settled, because the form decides where duplication falls.

Before a divisional structure is redrawn

Three figures are calculated before the decision: how many functions will be duplicated and what the duplication costs a year; what share of the smallest proposed division’s headcount support will take; and how much each division has to gain to recover the cost of its independence. The third is the figure that turns the decision from a preference about form into a judgement of what is reasonable.

Then the rule for allocating shared expenses is decided, written down and announced before the first year of the new structure, as a written decision rule of the kind formalization covers. A rule set after the first result has appeared is open to the reading that it was chosen to fit that result, and the first year’s figures then lose their value for comparison.

This is an explanation of the concept, not legal advice.

Related terms

• Decentralized Management
• Formalization
• Job Specialization
• Cost Center
• Cost Allocation