I first came across the Cynefin Framework when I was searching online for solutions to an all-too-common business problem: team members making decisions according to their preferred bias for action, skipping over contextual clues and data that indicate a different decision-making model is necessary.
This problem may show up in multiple ways, including:
A team member with a fondness for bureaucratic processes may believe every problem is a failure of process. The need, he believes, is to force the data into simple, pre-conceived categories and then act.
A team member with a background in logic-driven complicated systems, may see every problem as one needing deep domain expertise and sufficient resources for analyses. Analysis comes first, then sense-making, then response.
A creative thinker makes decisions quickly based on what worked elsewhere in the past, unintentionally reinforcing the status quo and raising obstacles to innovation.
Digital products, like the businesses that create them, inhabit dynamic, complex environments. These environments are made up of many agents acting on the system: trends, pricing, ever-changing algorithms, best-in-class events raising the competitive bar, tens of thousands and millions of consumers each with individual motives. A small change in the environment can yield an outsized and unpredictable outcome. What's often missing is an appropriate decision-making framework that accounts for complexity and uncertainty.
The Cynefin Framework is both a theory and a practice which helps teams discover what context/domain a problem/opportunity resides in, so that they can choose a domain-appropriate decision-making model. The premise is that you should think, analyze, and act differently dependent upon the domain. This is in contrast to a "one-size-fits-all" approach which Snowden rightly contends has been a tradition of management theory.
The framework was created by Dave Snowden, founder and chief scientific officer of Cognitive Edge. Snowden has developed innovative approaches to strategy and decision making which account for the requirements of operating effectively in complex environments. "Making sense of complexity in order to act," is his organization's guiding mission.
After studying the introductory online materials on the framework I attended a Cyenfin Framework workshop in New York City in April of 2015. Led by engineer-consultant Michael Cheveldave, the workshop was scheduled as a carry-over to LeanUX, and attracted an interesting mix of individuals from DevOps, marketing, academia, among other fields.
As you read further, keep in mind these two points underscored by Snowden:
The Cynefin Framework is a sense-making model, not a categorization model. As with Avinash Kaushik's models, it is critical to understand the thinking behind the model. This is not a "fill-in-the-quadrant" matrix. "In a sense-making model, the framework emerges from the data, while in categorization the framework is pre-given," explains Snowden.
Social context is key when applying Cynefin within a work environment. This article provides an elementary introduction to the framework, but to truly understand how a practical instance of it emerges within a social context I recommend attending a Cognitive Edge workshop.
To understand the framework, first consider its building blocks. When constructing Cynefin, Snowden drew upon three basic systems—complex systems, ordered systems, and chaotic systems.
Order was split into two domains: complicated and simple:
A fifth domain was added: Disorder. Disorder is the state of not knowing which domain your problem/opportunity resides in and/or which domain you are operating in. The outcome is that your decision model will likely not match the situation at hand. Snowden believes business decision-making is most commonly made in this domain of disorder.
The 5 domains
THE SIMPLE DOMAIN
Decision model: Sense → Categorize → Respond
The domain of best practice
Constraint type: Rigid
This is an ordered system. Cause and effect relationships within this domain are obvious, predictable, repeatable, and can be determined in advance.
Most situations that reside in the simple domain are infrequently subjected to change. Many of these situations are established processes. The domain is characterized by stability. Problems in this context, well-assessed, lead to straightforward actions.
The decision model is: sense, categorize, respond. “We see what’s coming in, we make it fit previously determined categories, we decide what to do," explains Snowden.
An example: a firm encounters an accounting problem. Perhaps some services were not being coded correctly. Best practice (in theory, regulation, and in software) is already well-established. The answer is obvious and instituted quickly.
Because the "right" answer is apparent to most, it is usually undisputed. Extensive communication is therefore typically not necessary.
The exception: Snowden underscores the importance of listening for weak signals in the Simple domain, especially when new systems are introduced. The feedback mechanism can be questioning and listening carefully to employees closest to the related day-to-day operations. Failure to detect weak signals can be "catastrophic."
More points from Snowden:
- "Since both managers and employees have access to the information necessary for dealing with the situation in this domain, a command and control style for setting parameters works best."
- Decisions in this domain can be easily delegated
- Many functions in the simple domain can be automated
- Best practice and process re-engineering makes sense here
Caution: Best practice is "legitimate in this domain" but "illegitimate in other domains" Snowden explains. This will become apparent as we explore the other domains. Consider: "best practice is, by definition, past practice," a point driven home in Snowden's classic "A Leader's Framework for Decision Making." (Harvard Business Review, David Snowden and Mary Boone, 2007).
THE COMPLICATED DOMAIN
Decision model: Probe → Sense → Respond
The domain of good practice
Constraint type: Governing
Situations that inhabit the complicated domain do not exhibit an obvious relationship between cause and effect; the relationship is there but it requires expert analysis to discover. The situation either requires the identification and deployment of an appropriate analytical method or bringing in outside expertise.
The decision model therefore is sense, analyze, respond.
“Accordingly I don’t apply best practice,” Snowden explains, “I apply good practice.”
Snowden underscores the distinction between “best practice” and “good practice.” “In a complicated domain there are multiple legitimate and effective approaches. Snowden cautions against trying to force one sole approach and the single “right” response.
Caution: Experts who primarily work in the complicated domain may err in judging all 'non-simple' problems as belonging in this domain. Or they may subject problems from the simple domain to expert analysis, delaying and over-complicating resolution.
THE COMPLEX DOMAIN
Decision model: Probe → Sense → Respond
The domain of emergent practice
Constraint type: Enabling
The complex domain harbors systems and situations without clear cause and effect patterns. Outcomes of actions are not predictable. An example might be pushing a new digital product (such as a website) or campaign out to tens of thousands or millions of people: you can't be certain in advance of what the outcome will be. Experimentation via a series of safe-to-fail experiments is critical.
The environment is complex: thousands, hundreds of thousands, or millions of factors (algorithms, disruptive events, individuals) interacting to produce often surprising outcomes. “It’s a system of light constraints on agents. Agents modify the system," Snowden states. The system evolves.
The decision model here is probe (collect data, learn), sense, respond. Conduct safe-to-fail experiments (cheap). Snowden: “We conduct safe-fail experiments, we don’t do fail-safe design.”
We see an example of that by looking at the results of A/B and multivariate testing. For example: a consultant was called in to help Skype design a new landing page. He created a draft based on best practice, but asked employee teams to create two more designs. When the three landing pages were tested, the one designed by the "expert" was not preferred by the customer community. Because they tested (probed), they avoided the brittle outcomes that may have come from deciding up front what customers "should" like. Test and learn frameworks within digital marketing are an example of Probe → Sense → Respond. (For an example of a good test and learn framework see Clearhead's Problem-Solution Mapping Methodology used for digital optimization).
As with all probing and testing, amplifying what is working and dampening what isn't is an essential part of the Probe → Sense → Respond cycle.
Activity in the complex domain yields emergent order and emergent practice, a new way of doing things.
Note: How do the complex and complicated domains differ? Martin Burns has a nice summary of the co-evolution dynamic that characterizes complex adaptive systems: "A complex system interacts with its context and evolves as a result, and a complicated system does not interact with its context or evolve.”
THE CHAOTIC DOMAIN
Decision model: Act → Sense → Respond
The domain of novel practice
Absence of constraints
In this domain no cause and effect relationship can be determined. "If we enter the domain deliberately it’s for innovation, but if we enter it accidentally then we need to stabilize the position quickly," explains Snowden. "We move very quickly, any practice will be completely novel."
In most depictions of the Cynefin Framework, the border between the simple and chaotic domains is depicted differently than the other borders. It may appear as a three-dimensional "cliff" for example. When success breeds complacency and leaders treat complex and complicated issues as belonging in the simple domain, a catastrophic fall into the chaotic domain may occur. Complacent firms rendered obsolete by disruptive innovations are another example that Snowden provides of this potential "cliff plunge."
"All the other boundaries allow for transitions. This: you fall over the edge and recovery is very, very expensive," explains Snowden.
"It therefore follows that you should manage primarily in the complicated and complex domains, and only move a small amount of material into the simple domain, because that’s actually highly vulnerable to rapid or accelerated change," he adds.
The central space of disorder signifies the state of not knowing which domain you are operating in. Snowden believes that most business decision-making actually occurs within this disordered space.
"The problem is we will interpret the situation according to our personal preference for action," he explains. "So what you get in a normal decision environment is that people are in a disordered space, assessing the situation according to their preference for action." Entrained thinking, organizational habits, personal action preferences (such as subjecting all/most problems to bureaucratic processes or expert analysis), as well as personal comfort levels, may dominate and guide decision making.
The alternative is to gather more information to ascertain which domain the problem/opportunity actually resides in, then approach it with a domain-appropriate decision-making model.
Snowden is Chief Scientific Officer at Cognitive Edge. Online and in-person training in the Cynefin Framework is available via Cognitive Edge.
Do You Want Dirty or Clean Consulting? (Richard Clayton)
Clayton argues against consulting trends based on Frederick Taylor's "one best way" delivered by "first-class men" and focused exclusively on "best practices."