Two Warnings about the Divisional Form of Structure

17 November 2022

Here are two excerpts about the Divisional Form of structure from my new book Understanding Organizations…Finally! that will be out in February. (It can be pre-ordered on Berrett-Koehler or on Amazon.)


Does Divisionalization Constitute Decentralization?

The prevalent answer to the question above has long been yes. Be careful about this.

The Divisional Form rose to prominence early in the Twentieth Century. As large companies that were structured functionally (purchasing, manufacturing, sales, etc.) began to diversify into other businesses, they ran into difficulties trying to coordinate these functions across the different businesses. Restructuring to the Divisional Form of structure, with one division designated for each business, whose management could control all of its functions, made coordination much easier.

But does this constitute decentralization? Sure, compared with keeping most of the decision-making power with the central management at headquarters. But what if the decision-making power delegated to each of the divisions remains largely in the hands of its own senior manager? Within the division, this constitutes centralization. And likewise across a whole company: with many thousands of employees, much of the power is concentrated in the hands of a few division managers.

This was evident in the most famous example of divisionalization, mistakenly labeled decentralized by the chief who introduced this structure. In 1908-1910, William C. Durant brought together a number of independent automobile companies—Buick, Cadillac, Pontiac, and others—to create General Motors, with a small central staff. When Alfred P. Sloan took over in 1923, recognizing the need to rein in the chiefs of these independent businesses, he instituted a divisional structure, with financial controls at the headquarters. In other words, Sloan did not decentralize General Motors: relatively speaking, he centralized it. But the word has often been misused ever since.

How ironic that while the Divisional Form used in corporate America came to be called decentralized, a similar structure used by the Communist countries of Eastern Europe to control their state-owned enterprises came to be called centralized in the West. Be careful about ill-chosen labels in an organization.


Driving the Divisions toward the Programmed Machine

It seems to be generally assumed that the Divisional Form can function with divisions of any form of structure. [In this book, the main forms discussed are the Personal Enterprise, the Programmed Machine, the Professional Assembly, and the Project Pioneer.]

Imagine if McDonald’s acquired Amazon. Don’t hold your breath, but if it did, no matter how different their businesses, with both companies being rather machine-like, that combination could conceivably work. Now imagine, instead, if McDonald’s acquired Apple. This would not be a happy union, especially for all those Apple engineers whose idea of innovation is not an Egg McMuffin. A Project Pioneer like Apple cannot function effectively under the control of a Programmed Machine, with headquarters’ analysts descending on these engineers in search of efficiency at the expense of creativity. This needs slack, not a straightjacket. The ups and downs associated with project work can also be anathema to a headquarters that expects steady increases in performance.

How, then, to manage those businesses in a diversified company that are less mass production, more project-like? Shortly after Jack Welch became CEO of General Electric, one of his advisers, who had read an early version of this book, called me with this question. What to do, for example, with the maker of jet engines compared with light bulbs? Keep the headquarters’ technocrats at bay, I suggested, by designating these divisions as Welch’s own preserves. In effect, use more of a Personal Enterprise at HQ to protect these particular divisions.

This was illustrated poignantly in another experience I had, with a British company called Thorn EMI, a conglomerate that combined Jules Thorn’s original lighting business with the EMI Music company and another business. An executive there told me that “Since Thorn died, nobody knows how to run this place.” Thorn would seem to have run it, not as a classic Divisional Form, but more like his Personal Enterprise. Could it be that the only way to run a conglomerate successfully, at least for a time, is to have an astute CEO who picks the right businesses and the right people to manage them? Period.

To conclude, in its conglomerate version, the Divisional Form can be described as a structure on the edge of a cliff. Taking one step ahead, it shatters into pieces on the rocks below. Behind are the safer places of an intermediate, or hybrid version of the Divisional Form (related-product, by-product, geographic), with some synergy across the divisions and fewer measurements to control them?