top of page
  • Photo du rédacteurErwan Hernot

Is Agile Decision Making Possible In Large Organisations?

Key idea: companies talk a lot about Agile. But can they walk the talk? The answer is probably : no except if they undergo a huge transformation. Training people in Agile is cool and SCRUM certifications for project managers are fine, but it's not the point.

Thanks to digital, big companies are now in a VUCA (1) world. For coping with it, executives dream of becoming nimble in decision making. Is it possible in corporations of such a scale? For now, many of them confronted this external complex first by … growing their organizational complexity, perhaps not the most relevant move towards Agile.

  • Structures are getting more complex: from strong product, functional, and regional axes to expansion of procedures for reporting, requiring evermore coordination bodies, meetings, and report writing. Oh and adding more specialised jobs than ever (can you understand what your friends do when reading their title on their business card?) This internal complexity relies on a strong assumption. It is a system you can handle because you are able to separate it into smallest subsystems. Once you’ve identified the subsystems, you can isolate them to see how they work and what part they play in the big system and build your understanding from the ground up. This approach to complexity works well but only down to a point. Organisations behave in ways that cannot be totally predicted by deconstructing them and studying the parts. Too complex: too many parts and interactions! As a result, the system as a whole determines in an important way how the system behaves. We are back in the unknown that internal complexity was supposed to suppress.

  • Hierarchy isolation. The model of complex organization fosters hierarchical "isolation". Leaders must acknowledge that there is a great remove between top management and those further down the organizational chart of the traditional Command and Control system. Problem is the latter are those engaged in the actual activities which a company is dedicated to. Consequently, this gap 1. instilled through years a tendency of executives to distrust the experience judgment of those under them. 2. It makes comprehension very difficult between the 2 groups. Not a very good start for Agile decisions.

  • Communication overload. Organizational complexity generates frictions between the different subsystems driven by different purposes. Communication is supposed to lessen these frictions. Fortunately (or maybe not?) optimistic would say "digital technology drastically reduced cost of communications". As a result, it has invited more people to the party aka into the communication flow via email, Slack, and internal knowledge-sharing platforms. The first visible result is too many meetings and email threads: with all that time spent in reporting and coordinating, telling and “selling” your opinion has become the job. There is a little time left for actually doing ;) Executives face then to a great degree, a cognitive constraint that confronts all of us: making decisions despite having limited time and ability to deal with information overload. Human minds prefer causality over complexity. Indeed, the risk is that they tend to linear thinking and see the world as a place full of explainable events with simple causes and simple effects.

  • Relying too much on metrics to handle this communication overload. Based on this rationale, metrics are therefore a tempting means of dealing with this bounded rationality. Relying only on metrics to lead has serious limitations though: this hard information is often limited in scope, lacking richness and often failing to encompass important non-economic and non quantitative factors. Even more: much hard information is top aggregated. Problem is that a great deal of information is lost in such aggregating. How much does the bottom line tell about the condition of a company? What do sales numbers in the CRM (2) reveal about the quality of relations between sales teams and customers for an analyst who hasn’t see a customer in his entire life ?

  • This corporate context doesn't favor Agile mindsets (think simple, think client first, think added value rather than defending your territory). Rather it cultivates narrow mindsets with players trying to game the rules, process and structures for protecting their autonomy.

The result, in these hyper connected organizations is a paradoxical decrease in high-quality dialogue leading to sound basis for decisions. Leaders are less able to delegate decisions cleanly, and the number of decision makers, at the wrong levels, has risen...growing complexity. Is Agile decision making possible in such a context? It could be if things are done in the proper order. Don't start with Agile trainings if you are still in a Command & Control landscape. Executives should start by rethinking the organization and re-engineering their decision making process based on the following principles:

  1. Put the organization upside down. (Not kidding)

  2. Share the data to create knowledge.

  3. Adapt the decision making process to the situation at hand.

  • Put the organization upside down. Don't change the approach system/subsystems though. Transcend it wtih crystal clear principles. The boldest changes put the customer experience at the center of the structure and re-organize what evolves around it (3). This revolution goes as far as create 3 major components structuring a company. For instance, customer experiences teams (C1) which subdivides into smaller teams like “Browse the different products families “ (C2) to eventually smaller units like “Choose a payment method” (C3). Business processes teams and Technology systems teams share the same subdivisions. These components mix different professionals rather than siloing them in a traditional hierarchical organizational chart. Thinking in such organizational terms drive all the players closer to real problems and makes it easy to spot constraints and take decisions... if they are able to cope with the diversity of thinking coming from different jobs involved. This a profound and daring change ! It impacts silos, territories, hence powers. Can big corporations make the jump? One can doubt knowing that when they buy start- ups, theirs leaders lucidly keep them at good distance to PROTECT them from their own operating way! Let's pretend it is possible though, for the sake of this post ;)

  • The solution for many organizations looking to untangle their decision making is - not surprisingly - to become flatter. For better overall control of the complex system, most of the knowledge and decisions should be made in the subsystems. It is a smart way to override the cognitive constraint from the communication overload. Executives push decisions and responsibilities down to a level where someone has a meaningfull information because it is smaller in size and more accurate that the one aggregated at the top. These executives understand that they must make as few decisions as possible. Strategies must be more openly discussed and executives must listen to dissent voices. In order to do that, they can empower people with open access to data (4). Accessing to all the data gives, to those down the trench, a power of asking questions and formulating competitive arguments and creating knowledge relevant to take the best suited decisions. This empowerment supposes people trained in soft skills. Take experts for instance. Their participation adds value only if each expert gives credit to the opinion of others…This empowerment needs also people trained in handling data. After all, you train your algorithms with data ; why not your people? Put this managers ‘empowerment on steroids with AI ;) AI doesn’t get overwhelmed by the size and complexity of information the way we humans do. Swelling stockpiles of data, advanced analytics, and intelligent algorithms are providing managers with powerful new inputs and methods for making all manner of decisions.

  • For executives and managers, decisions just happen without thinking more about the very process. Re engineer the decision making process to fit the situation. And to reassure the empowered people possibly afraid of the accountability coming with the decision: there are methods they can rely on. For instance, we know that there are, mathematically, two ways to reduce the collective error: to reduce individual errors, or to increase the variance of the estimates. We tended to use the first option before the digital age: we reduced individual errors by restricting the numbers of decision makers and choosing the best and brightest. This option doesn't stand in the VUCA world : the leader who knows everything is a complete illusion and the expert who knows a lot has just a narrow view of the problem to solve. Consequently, one can increase collective expertise either by recruiting experts from different professional background or by calling in more people (in the meaning of "the crowd"), aka multiplying points of view. See table below freely inspired by Samantha Slade, "Going Horizontal", Berrett-Koehler Publishers, 2018.

Despite the digital era, Agile decision making seems eerily exotic in large organizations. It will come though if executives provide the following conditions : the power to decide is clearly delegated to those on the ground ; they are trained to use data and AI and they understand that before taking a decision, you need to engineer it depending on your situation. Executives then are more attentive to the making of the decision than to the decision itself. From decision makers to decision facilitators, it could be a long way for some leaders...

(1) VUCA: volatile, uncertain, complex, ambiguous.

(2) CRM:Customer Relationship Management.

(3) Agile At Scale, Harvard Business Review May June 2018 Darrel K. Rigby, Jeff Sutherland, Andy Noble. But they didn't invent anything. Organizational sociologists like François Dupuy in France use the concept since the 1990ies.

(4) "Data" refers to "external" data (business metrics), "internal" data (operations metrics). This is obviously not confidential data (R&D for example).


bottom of page