When they are run as mere organisational projects, the centralisation or decentralisation of a company most often comes up against human obstacles, all too human… Yet they are among the most common changes, whether they are owned as such or are the consequence of other projects.
Let me say first of all that I do not believe there is such a thing as an ideal organisation. If only because, if there were, it would be known and promoted by the big consulting firms, and would have prevailed years ago.
My second conviction, on this subject, is that what matters most lies in the implementation. Defining a strategy is one thing, and it can be very “clever”. But if it is not implemented, it is useless.
Centralise or decentralise: a matter of power?
More than others, but like many others, French organisations are very sensitive to the question of power. It is striking, in fact, when you follow a programme abroad on change in organisations, as I have done twice: the only topic on which French authors are cited is power.
This “French school” also has a very particular (and reductive) slant: that of “power over”. Power in the tradition of Max Weber: that of organisations and their bureaucracy. Or that of “domination”, whether individual or collective, in the footsteps of the baleful Foucault.
And even if we free ourselves from this sad French passion, we find in Crozier and Friedberg, who are infinitely more interesting and commendable, the same interest in power, whether that of individuals or of organisations.
That, in my view, is one of the great weaknesses of our French political (and therefore sociological) tradition, which pushes towards conflict through the confrontation of powers. We do not look for consensus, we set collective play aside, we think in terms of winners and losers.
What we miss, with this monolithic approach to “power over” and power relations, is the possibility of thinking about “power with” and “power to…”, to take up the distinction drawn as early as the 1920s by Mary Parker Follett, the American pioneer of management thinking. For better, and of course also for worse, all proportions kept: Hannah Arendt, for whom power is born of acting in concert, also showed, in Eichmann in Jerusalem, how collective momentum, when everyone takes part without thinking about what they are doing, could lead ordinary individuals to take part, mechanically, in the worst mass crimes.
Above all, we forget the central question of the balance of powers, dear to the English-speaking tradition, and of the virtuous confrontation of interests in the service of a shared objective. An irony of history: it was a Frenchman, Montesquieu, who gave it its theory (by observing, it is true, the English constitution), but it is the English-speaking world that turned it into a practice. And so we deprive ourselves of considering and exercising other ways of relating, other ways of acting.
At the heart of the matrix
A striking illustration of the French difficulty with this balance of powers can be found in so-called matrix organisations.
Let us recall what they are: organisations in which two lines of authority cross, most often a “functional” axis (areas of expertise, functions) and a “business” axis (products, projects, countries or customers). Where they cross, each person reports to two managers. The matrix therefore chooses neither centralisation nor decentralisation: it holds the two in tension, and forces people to decide together.
These organisations, and the ways of working that follow from them, are often rejected in France in the name of a “French exception”. Yet they have one merit: they offer a perfect picture of the meeting between the so-called operational entities (those of the “periphery”) and the so-called support entities (those of the “centre”).
This customary distinction between centre and periphery also reflects a certain myth of the worker, inherited from the political lens of “class struggle”: the producer brings in money, while the others are cost centres. Which always leaves me wondering. For if support functions were nothing but cost centres, companies that are always keen to cut their costs would have got rid of them long ago, legal and regulatory obligations aside (and in France, those weigh heavily…).
In the case that concerns us here, the matrix gives concrete form to the need for production entities to take the expertise of central functions into account, or to draw on it (which is a more positive way of putting it).
In the confrontational approach, one of the two axes wins: the one that the company’s strategy, culture or practices value. In a collaborative approach, the experts become what are known as “business partners”: people who help to find solutions, or to improve results by making the work of production easier.
In terms of power, a strategy of centralisation officially gives power and its resources to the centre: mostly central functions, bodies of expertise and/or of “governance” (what an ugly word).
Conversely, a strategy of decentralisation tends to give power to the field, to the “business units”, which are made more autonomous, and more accountable too. For this decentralisation is first and foremost about decision-making: a prerequisite, in principle, for decentralising shared resources – expertise in particular, which in the centralisation scenario is on the contrary pooled and shared but, often too, placed in a position of imposing its views.
The power to decide?
So, while my preference for a virtuous balance of tensions and for collaborative work is easy to guess, there is no hiding the weak points of these forms of organisation either.
Many people criticise them for making certain decisions impossible, when stakes and points of view collide.
To resolve this tension rather than live with it – particularly when everyone digs in – some companies have found a “solution”: procedures!
Carefully drawn up, these describe, in theory, prerogatives and responsibilities… They are the “tablets of the law” that designate the holders of power and describe their instruments.
They reassure some, but deprive all stakeholders of the expected benefits of the virtuous tension of a balance of powers, which aims to find a solution together by reconciling constraints and objectives.
I remember an assignment in a large energy production company. I had been brought in to resolve a conflict between processes: a conflict identified as critical by the internal inspectorate and which had dragged on, to the detriment of the operational efficiency of the entity concerned. The activity as a whole had been described by a series of processes which, inevitably, cut across one another. The most salient conflict, to my mind, was the one between the “produce” and “maintain” processes. Because when you produce, you use machines and people, and you “wear them out”. And when you maintain, you no longer produce.
One of the people I was dealing with, a more than brilliant engineer, had proposed creating a new process that would settle this conflict: a “decision machine” that would rule in favour of one or the other, according to criteria that would need to be defined.
Transferring the power to decide to the designer of the algorithm, then, rather than living with the complexity of human interactions: now there is a very topical debate… But one which, in practice, obviously does no more than move the battlefield from the daily life of the organisation to the lines of code.
Other organisations deliberately choose ambiguity and vagueness, including in their procedures. Precisely because they are looking for constructive tension, consensus or compromise: a solution that may not be ideal but will reconcile opposites and, above all, bring everyone on board. I will give an example further on, with a matrix organisation designed in that spirit.
Crisis, a borderline case
There remains one situation that some describe as absolutely requiring the centralisation of the organisation and of decisions: a crisis. A borderline case, which deserves a closer look.
This plea for centralisation in “exceptional” circumstances owes more, in my view, to the authoritarian temptation of its advocates than to a search for effectiveness. For while an emergency justifies taking immediate decisions, in a time frame that the search for compromise does not have (you will not try to persuade a child who is running across the road after a ball; you will shout a warning, or even hold them back), a crisis requires you to take account of both its complexity and its duration. Playing as a team in order to understand better, enriching the decision in order to guarantee its implementation: that is the condition for “being” – being present, clear-headed and followed – and for lasting, by preserving, physically and intellectually, your capacity to analyse and decide.
Some advocates of centralisation in a crisis, incidentally, invoke a military organisation that they fantasise about and often know little of. For while in combat you do not challenge the leader’s orders – there is no time – the preparation of the operation is collective (take, for example, the courses on MEDOT, the French Army’s method for developing a tactical operational decision, run by its War College, the École de guerre – Terre).
And execution itself relies on a way of working that calls on initiative and decentralised responsibility: what General Pierre Schill, Chief of Staff of the French Army, has described with the explicit phrase “command by intent”, known in English as mission command. So even the army seeks balance: the intent is set at the centre, and the way to achieve it is decided in the field.
Beyond power, “culture”
Beyond this question of decision-making power, the choice of a form of organisation, and above all its implementation, must take other factors into account.
There is, in particular, “culture”. That of the company, that of the country: I mentioned French particularities above, but there are others, and they all meet in multinational organisations or, quite simply, in those that recruit beyond a single socio-cultural group.
This cultural approach is often disparaged, even denounced, because it can verge on simplification and caricature. And indeed, a reductionist approach overlooks the differences between individuals that arise from many factors over the course of a person’s life. And over the life of organisations, too, which are born, grow, live, evolve and sometimes disappear, absorbed by others or dissolved.
But there are very real fundamentals, whether they are “national”, for example, or linked to professions. It is easy to see that the culture of a sales function is not the same as that of a legal department. Or that a company of engineers does not have the same collective reflexes as an assembly of sociologists.
And, to come back to matrix organisations, I remember the deep dismay of a German-speaking Swiss engineer faced with the tensions and uncertainties born of the new organisation that had been put in place. It was a matrix organisation not in two dimensions but in three (products, countries, functions), in which the “boss” was not at the centre but at the periphery… Designed precisely to lead stakeholders to find common ground without readily resorting to hierarchical arbitration, it provoked in him both incomprehension and revolt. And while, obviously, a person cannot be reduced to their nationality or their profession, looking to these influences helps us to imagine, in broad strokes, what he was going through.
Every organisation has a culture: it is expressed, with its sometimes conflicting variants, at every level of the organisation. But this culture, its habits and above all what goes unsaid (what Edgar Schein calls its “basic underlying assumptions”, and which are the great difficulty of working on corporate cultures), has been built up over the years. It has “encoded” the people who arrived and who reproduced, often without realising it, its fundamentals. For those who stay adopt it, while those who do not fit exclude themselves from the group.
Deciding to centralise or decentralise an organisation therefore also means taking on its culture. The keel of the sailing boat: the part that gives the whole its stability when the sails are playing with contrary winds.
This does not mean that such a transformation is impossible. But that it can be slow, and above all that it must take many aspects into account, beyond decision-making alone. A “systemic” transformation, as consultants say.
Agile company or bureaucracy?
The centralisation or decentralisation of decision-making will be helped or, on the contrary, held back by the company’s culture when it comes to “bureaucracy”. Beyond the processes, which can be changed, are people more used to taking initiatives and resolving tensions through dialogue and negotiation, or to falling back on procedures, “decision machines” and control?
Decentralisation will shift responsibility for decisions and their implementation onto local players. Who may have neither the desire nor the skills to take it on. And it will take it away from central players, who may feel they are losing their social usefulness and the meaning of their work, unless they adapt to this change, for example by becoming facilitators and experts in their professional field.
And conversely, centralisation will take away local players’ room for initiative and manoeuvre, leading them, with this loss of power or quite simply of meaning at work, to leave the company, join the centre, or slip into a form of concealment of how things really work, or of disengagement. Just as it will place a heavier workload on the centre, which will probably lack both the skills suited to local particularities and the resources.
To the difficulties of cultural transformation will therefore be added the challenge of building skills: professional skills (“hard skills”) and interpersonal skills (“soft skills”). Areas of transformation that are often forgotten.
Implementation: evolution or break?
A transformation project that aims to centralise or decentralise an organisation is therefore a complex project. Humanly complex.
One of the questions to ask is whether it is a project of evolution or a project that makes a clean break: which, in terms of change management, is radically different.
In some contexts, and particularly when change is imposed from outside (a merger or acquisition, for instance), the transformation is far-reaching. It is then a breakthrough change, which must be prepared with care because, at one moment or another, you will move from system A to system B. Unlike a moderate change, which can be implemented more gradually.
In many cases, it is an evolution, an adjustment along the gradient between centralisation and decentralisation. That takes nothing away from the many factors to be taken into account in implementing this transformation. But it reduces its intensity. The challenge is perhaps not to avoid deadlock, but to ease and speed up the transformation. And to anchor it for the long term, until the next swing of the pendulum. For as I said in the introduction, there is no ideal organisation. And these organisational changes are sometimes due to exogenous factors, which are imposed on us. They are also, sometimes, driven by endogenous factors, intended to challenge bad habits and to seek, through the change they provoke, an overall improvement in performance.
That makes sense, but every transformation project goes through a “valley of tears”: the period during which, once the transformation is under way, overall performance drops, because of the necessary adjustments and the technical, organisational and human “bugs”. Well-run change management reduces the depth and width of this valley, and makes it possible to return to the previous level of performance, and then to exceed it (since that is the objective). But the valley will always remain a moment of destabilisation, and that is why a “centralisation / decentralisation” transformation can only be carried out once in a while.
Power struggle or deep transformation? Both, of course: you cannot move the power to decide without touching the culture, the skills and the habits of those who exercise it. That is why, as Montesquieu said of the laws that make for the stability of an institution, an organisation should only be touched “with a trembling hand”.
Translated by Claude (Anthropic’s AI) from the original French text.


Laisser un commentaire