Read Who Says Elephants Can't Dance?: Leading a Great Enterprise through Dramatic Change Online
Authors: Jr. Louis V. Gerstner
Tags: #Collins Business, #ISBN-13: 9780060523800
It isn’t a question of whether elephants can prevail over ants. It’s a question of whether a particular elephant can dance. If it can, the ants must leave the dance floor.
I don’t intend to describe here all the elements of creating a nim WHO SAYS ELEPHANTS CAN’T DANCE? / 243
ble, responsive, large enterprise.1 Certainly the matters just discussed—focus, execution, and leadership—apply to enterprises of all sizes.
There is one item, however, that I want to comment on, because it was essential in getting IBM dancing again. This is the issue of centralization v. decentralization in large enterprises.
A corollary of the “small is good, big is bad” mantra is the popular notion that, in large enterprises, decentralization is good and centralization is bad.
In the 1960s and 1970s McKinsey built a powerful reputation promulgating decentralization to corporations all over the world. It first pushed the idea in the United States, moved into Europe in the 1970s, and eventually went to Japan (where the idea was rejected emphatically by most Japanese companies).
Decentralization had a powerful intellectual underpinning, and over the course of a few decades it became the “theory of the case”
in almost every industrial and financial enterprise. The theory was very simple: “Move decision making closer to the customer to serve that customer better. Give decentralized managers control over everything they do so they can make decisions more quickly. Centralization is bad because it inevitably leads to slower decision making and second-guessing of the people on the firing line, closest to the customer. Big companies are inevitably slow and cumbersome; small companies are quick and responsive. Therefore, break big companies into the smallest pieces possible.”
There’s a lot to be said about the power of this construct, and it should, in my opinion, continue to play an important role in organizational behavior in large enterprises. However, I believe that in the 1980s and 1990s it was carried to an extreme in many companies, with
1This issue of entrepreneurial behavior in large corporations had been a passion of mine
for decades. See Harvard Business School cases on corporate entrepreneurship, based on
my activities at American Express. (Harvard Business School Cases 9-485-174 and 9-485-176, copyright © 1985).
244 / LOUIS V. GERSTNER, JR.
unproductive and, in many cases, highly disruptive results. Too often managers began to express the view that they lost their manhood or womanhood if they didn’t control
everything
that touched on their business. Consequently, every decentralized business had its own data processing center, human resources group, financial analysis team, planning organization, and so on. Decision making was, in fact, fast if the decision touched only on a single decentralized unit.
However, when multiple segments of the enterprise had to be involved, the highly decentralized model led to turf battles and inadequate customer responses because of incompatible systems in the bits and pieces of the enterprise.
Moreover, as long as profit margins were fat, the extra staffing may have been tolerable, but as we approached the raw-knuckled competition of the 1990s with capacity excesses in almost every industry, companies could ill afford duplicating staffs and process development at every level of the company.
Yet cost and speed are not the only issues. In many large institutions, the decentralized units were created for a different world or acquired as pieces of a larger mosaic. Now these companies are trying to create new value through the combination of historically separate entities. Examples abound all over the corporate world: financial services companies striving to create integrated offerings for customers from historically disparate product units; industrial companies trying to redefine their value to their customers as something more than a traditional product—usually a service wraparound; media companies trying to package advertising opportunities that combine various pieces of their enterprises; telecommunications companies trying to attract and hold customers through integrated offerings of voice, data, and entertainment.
This is not a challenge limited to the corporate world. University presidents have been struggling for decades to create interdepart-mental programs that integrate various fiefdoms of the academy.
Memorial Sloan-Kettering Cancer Center has been working for years WHO SAYS ELEPHANTS CAN’T DANCE? / 245
to create cross-departmental treatment protocols, i.e., an integrated approach to a particular type of cancer that combines surgery, chemotherapy, and radiology. Both in universities and medical centers this is hard work, because the department chairs who run the traditional decentralized units have enjoyed years of carefully guarded autonomy.
The problem of decentralization exists in government, too. The United States intelligence community is a hopeless hodgepodge of overlapping yet ferociously independent organizations. When a new threat arises (such as domestic terrorism), the task of redirecting the intelligence assets of the country away from the missions they were originally designed to carry out to meet the new challenge becomes an integration task of gigantic proportions.
Too Expensive, Too Slow
I believe that in today’s highly competitive, rapidly changing world, few if any large enterprises can pursue a strategy of total decentralization. It is simply too expensive and too slow when significant changes have to be made in the enterprise. Thus, what every CEO has to do is decide what is going to be uniquely local (decentralized) and what is going to be common in his or her enterprise. Note the absence of the word “centralized.” It is not a question of centralization v. decentralization. Great institutions balance common shared activities with highly localized, unique activities.
Shared activities usually fall into three categories. The first and easiest category involves leveraging the size of the enterprise. Included here would be unifying functions like data processing, data and voice networks, purchasing and basic HR systems, and real estate management. For the most part these are back-office functions that yield to economies of scale. It is absolutely foolish for a CEO to accept the whining of a division president who says, “I can’t run my business
246 / LOUIS V. GERSTNER, JR.
successfully without running my own data center, managing my own real estate, or purchasing my own supplies.” Even a company as diverse as General Electric effectively exploits its scale economics in back-office processes.
The second category involves business processes that are more closely linked to the marketplace and the customer. Here the drive to common systems can offer powerful benefits but most often involves linkages among the parts of a business that may or may not make sense.
I’m thinking here of common customer databases, common fulfillment systems, common parts numbering systems, and common customer relationship management systems that permit your customer-service people to provide integrated information about everything a customer does with your company.
On the surface it would seem that these are logical and powerful things to do in an enterprise. Nevertheless, they usually require profit-center managers to do something very hard—relinquish some of the control they have over how they run their business. Staff executives, consultants, or reengineering teams cannot do this without active line management involvement. The CEO and top management have got to be deeply involved, reach tough-minded conclusions, then ensure that those decisions are enforced and executed across the enterprise. It takes guts, it takes time, and it takes superb execution.
A Step Too Far
Having made the point that decentralization has gone too far in many institutions, I quickly add that there is a ditch on both sides of the road. My concern is that today many CEOs are seeking utopian levels of integration. This is the third—and most difficult—area of common activities, involving a shared approach to winning a marketplace, usually a new or redefined marketplace. These activities are
WHO SAYS ELEPHANTS CAN’T DANCE? / 247
difficult because they almost always demand that profit-center managers subjugate their own objectives for the greater good of the enterprise. As such, they can be enormously controversial inside a company and lead to bitter and protracted struggles.
Here’s an example: During my time at American Express I was running the so-called Travel Related Services business, which included the American Express Card division. It was the largest and most profitable segment of American Express. American Express bought a brokerage company as a step to create a one-stop financial supermarket. In the course of enticing the brokerage company to join American Express, the deal makers promised the brokerage that they would have access to the American Express cardmember list.
In other words, they would be allowed to make cold calls to cardmembers to try to sign them up for brokerage accounts. When this became known to the card division, there was an open revolt. Those of us who had built the card division believed it was assembled on a basis of trust, privacy, and personalized service. Cold calls from securities brokers did not fit into our definition of customer service.
The war went on for years, and the integration or synergy that the CEO had hoped to achieve not only never happened, but it led to the departure of many senior executives and ill will that contributed to the eventual disposition of the brokerage business.
It is very easy to conceive of how various units in a company can work together against a common enemy or seize new ground in a competitive industry. Think about all the financial supermarkets that have been constructed (and almost as many deconstructed).
Think about all the mergers and acquisitions that have taken place in the entertainment and media industries; The New York Times Company buying cable companies and sports magazines; Disney buying a television network; the behemoth known as AOL Time Warner.
How many times have we watched two CEOs stand up at a press conference and make claims about the extraordinary benefits that would be achieved once they merged their companies to create a 248 / LOUIS V. GERSTNER, JR.
unique combination that would bring new services and new benefits to the marketplace?
Well, we’ve all seen what happened in almost every one of these instances. They fail. Why? Because in most cases the CEO must ask people to do things that are inextricably and inexorably in conflict.
Divisions are asked to compete against their traditional competitors, focusing on maintaining a leadership position in their individual markets. At the same time, they’re asked to join with other divisions in their company in a much broader fight that inevitably involves giving up some resources or assets that are needed to win in their traditional market.
There is great risk in asking a decentralized unit of an enterprise to be good at its traditional mission and, at the same time, fulfill a shared role in creating value in a new mission. The conflicts—most often having to do with resource allocations, but also with pricing, branding, and distribution—will be overwhelming.
I am about to suggest something that will annoy almost all the world’s management consultants (they make a lot of money defining
“new industry models” and describing “synergy opportunities”): CEOs should not go to this third level of integration
unless it is absolutely necessary
.
For most enterprises the case for integration ends with category two. Category one is a no-brainer; most back-office functions can be combined with significant economics of scale. Category two (integration of “front office” functions that touch the marketplace) can produce significant benefits, but the integration must be executed superbly or the benefits will be decimated by the parochial interests of individual units. Category three is very much a bet-the-company proposition.
However, there will be times when a CEO feels it is absolutely necessary to bet the company on a new model—a truly integrated model (I recently met on two occasions with CEOs who run media and entertainment companies, and they said they are agonizing over this
WHO SAYS ELEPHANTS CAN’T DANCE? / 249
decision). If you decide to go down this path (as I did at IBM), let me outline a few of the steps that I think are critical to making a successful conversion. I can’t and won’t belabor this issue here; it would fill another volume. What follows are only very introductory comments.
Shift the Power
One of the most surprising (and depressing) things I have learned about large organizations is the extent to which individual parts of an enterprise behave in an unsupportive and competitive way toward other parts of the organization. It is not isolated or aberrant behavior. It exists everywhere—in companies, universities, and certainly in governments. Individuals and departments (agencies, faculties, whatever they are called) jealously protect their prerogatives, their autonomy, and their turf.
Consequently, if a leader wants fundamentally to shift the focus of an institution, he or she must take power away from the existing
“barons” and bestow it publicly on the new barons. Admonishments of “play together, children” sometimes work on the playground; they never work in a large enterprise.
At IBM, to be a truly integrated company, we needed to organize our resources around customers, not products or geographies.
However, the geographic and product chieftains “owned” all of the resources. Nothing would have changed (except polite platitudes and timely head nodding) if we didn’t redirect the levers of power.
This meant making changes in who controlled the budgets, who signed off on employees’ salary increases and bonuses, and who made the final decisions on pricing and investments. We virtually ripped this power from the hands of some and gave it to others.
If a CEO thinks he or she is redirecting or reintegrating an enterprise but doesn’t distribute the basic levels of power (in effect, redefining who “calls the shots”), the CEO is trying to push string up a hill.