Uncertainty and fluidity are everywhere. Any fields, industries, sectors are registering increasing levels of complexity. Organizations are learning that agility help them to remain attentive and adaptive to the context they are in, leveraging on any emerging opportunities and avoiding do be caught in trap by unforeseen risks. How is that possible? Agile relies on the same three pillars (Transparency, Inspection, Adaptation) of the scientific approach, which relies, in turn, on experiments to validate every and each hypothesis. In this article we will explore how experiments can help organization to navigate uncertainty and accelerate the production of value.

Doctor Ralph D. Stacey is one of the pioneers who studied the implications of the natural sciences of complexity for understanding human organisations and their management.

Through his matrix (see above), he greatly described how we can categorize systems, according to the agreement on what to build (how much we understand of the system that must be built and how much we are in agreement?) and the certainty on how to build it (do we have the necessary competences, skills, technology and knowledge to build it?).

According to the answers we give to those questions, he identified four situations/systems we could find ourselves: Clear (or Simple), Complicated, Complex or Chaotic.

Dave Snowden brought this one level next with his Cynefin framework, which describes what different behaviors, decisions and actions we could take when facing each of those systems.

Most of the contexts we live are complex ones, thus it worths to shed a light on this very quadrant of the framework. For a description of all quadrants see one of my previous posts here.

When we find ourselves in complexity, there are no predefined solutions or approaches and we need to proceed by making hypothesisexploring the context, working incrementally and gathering any emerging feedback from the fields that can confirm, or refute our assumptions.

The suggested approach is probe-sense-respond which means that the first thing we must do is to try something, doing some experiments in order to sense any emerging pattern, learn from that and the respond accordingly.


Complex systems are dispositional in nature.

That means that they are oriented and disposed to “move” in a certain direction, which we want to discover in order to leverage on that. They evolve in certain ways, but causality is not known or determinable upfront, thus the only way to understand it is to make experiments.

Cynefin use the term probe to stress the fact that we want to literally “put” probes (experiments Ed.) into the systems and learn how the system reacts.

These probes are small experiments that must be small ones and, moreover, safe ones; the latter means that if we fail (and in complexity the percentage of failure is very high) we can afford it, without compromising the whole project, program, product or service development activity.

Cynefin suggests to build a portfolio of probes which “only” purpose is not to decide what to do, but however to clarify the situation by maximizing the learning and, completely in alignment with the scientific approach, to try to invalidate our thesis or ideas.

If they remain alive after the execution of these experiments, could mean we are on the right spot.

Our portfolio should consider several parallel experiments that directly, or even indirectly, should bring a piece of learning.

The Lean Change Management Cycle

Lean Change Management (LCM) is the Lean-Agile way of introducing change in organizations. It is rooted on Lean Startup approach and diversify itself from standard change management approaches by totally embracing agile and lean values and principles, which suggest to:

  1. proceed empirically,
  2. always taking individuals and their interaction and feelings in great account,
  3. actively engaging involved people to co-create change,
  4. visualizing change management initiatives and insights and
  5. managing resistance to change is not something to manage but, rather, to understand and treat as any other feedback.

Getting back to complex systems, we can “steal” one great LCM’s tool (and relative process) that represent how to manage portfolio of experiments and their life cycle.

No alt text provided for this image

The LCM cycle is as simple as it is powerful. Everything starts with insights we are learning and gathering from the context we are exploring. These insights let us make some hypothesis on which to formulate assumptions.

Based on the latter, we should identify some candidate experiments aiming to demonstrate that our assumption is wrong (remember the scientists and their approach).

These candidate experiments are then assessed against a Cost/Value matrix as follows:


Now you should have a clear prioritization on what experiment to launch first. Get ready to activate the relative Prepare>Introduce>Review sub-process:

  • Prepare: the selected experiments need some planning. It is necessary to clarify what is the piece of learning you want to get out of it, what resources you need, what is the time necessary to execute and who will be the recipient (people) and who will be impacted.
  • Introduce: the experiment is executed.
  • Review: results are inspected, insights are extracted and will feed the next outer cycle.

As written above, experiments should be small enough to be run in a few days and in parallel with others. One last Lean principle we need to take into account, is to visualize what you are executing and what information and patterns are emerging, to allow everyone involved to reason on that and to formulate new informed hypothesis.

A Trello board, in time of Covid-19, is a great tool to keep everything visible, transparent and accessible to everyone:

Exploring uncertain context could be a tiring and stressing task, but proceeding empirically by running incremental experiments, will give infrared rays to see through the fog and definitely delving into complexity.