10 steps to resolve complex problems

Regardless of whether you follow Drucker, Smalley, Senge, or other modern thinkers, there are really only three types of problems: simple, complicated, and complex. Almost everyone understands what a simple problem is. It is routine, has a few parts, and implementation of system remediation straightforward. However, few people are familiar with the difference between complicated and complex problems. Complicated problems have multiple parts and it requires time and effort to understand how the system is working to resolve them. Complex problems also have many parts, but even if you understand how they interrelate within the system, you will not be able to predict their behavior.

Over the years, organizations have developed expertise to resolve simple and complicated problems but still struggle to resolve complex problems. This post will provide you with ten simple steps to resolve complex problems.

1) Set Scope

Organizations are open systems. They are created from nested groups that depend on each other. It is difficult to draw a boundary line within a certain group of an organization just as it is difficult to draw the boundaries of an organization. 

When it comes to solving complex problems though, a clear scope of the problem, with clear boundaries, must be in place. Without clear boundaries, the problem’s resolution will result in an enormous effort with no results. Defining the problem scope or domain is a critical component of any serious effort. It is important to keep in mind though that scope typically changes as the problem is better understood

2) List all parts

After identifying the scope and boundary, the next step is to list all the “parts” that compose the defined scope. Then you must define the level of the parts. Using too many parts from lower levels will create clutter, too many high-level parts won’t bring value or the results you are looking for.

Problem domain parts should include every autonomous element in the system which makes decisions, takes actions, as well policies, procedures, or rules that guide autonomous parts. In most cases, these include roles, people, and groups. If you can focus on roles rather than individuals, your results will be clearer.

Usually, parts of the problem domain will also include technologies, applications, or processes used by functions, people, and groups. A list of all the parts, ordered by their type, reveals a refined scope of the complex problem under investigation.

3) Understand how parts interact

After all the parts are identified and listed, we want to understand how they interact. All complex problems have nonlinear interactions by nature. Attempting to describe nonlinear interactions from a complex problem, linearly, creates an obstacle that will prevent you from gaining any substantive understanding of the system. There are three types of interactions though, which once captured, will provide a significant understanding of the problem: impacts, flows, and depictions of the interactions.

First, you must understand how each element impacts each other. Impacts are nonlinear. For example, if A impacts B, then B impacts A as well. The level of effort A put, not necessarily define the level of the results. Our goal is to capture how all parts impact each other and if the impact is positive or negative.

Second is to understand how the information and materials flow between parts. As with impacts, flows need to be depicted non-linearly.

Third is to depict how interactions between people or roles (rules they are following based on available data) shape the aggregate behavior of groups.

4) Find flaws in the system that connects all the parts

After gaining an understanding of the complex problem we want to fix, the effort should shift to finding problems and opportunities in the way the system, that supports the problem domain, is set up. This is not a focus on business applications, but a method to identify the problems in the management system (policies, procedures, rules, norms, processes, structure, flows, etc.) that connect all the parts.

Utilizing the previous three steps, multiple tools manifest to identify opportunities in how parts are connected. This effort should focus on identifying the problems within the management system and how to resolve them.

The first focus should be on the management system. If there are discoveries related to the management system, do not dive into groups and roles at this time. Usually resolving management issues will change all the interactions in the system which creates new opportunities. If no flaws are found in the management system, then focus on groups and roles. Try to connect what their contribution to the problem is.

5) If the parts are human, understand their human drive

After uncovering and understanding opportunities that relate solely to the way the management system was set up, the focus should move to the human factor. In this step, the goal is to understand what drives people’s behavior and actions.

Using mental models as a tool allows us to gain insight into what is the human drive affecting the system. Noted British statistician George Box said, “Essentially, all models are wrong, but some are useful.” Although simplifications, they can be useful to obtain insight. Mental models can be defined as how people think during certain events, how they cooperate or compete for scarce resources, and narrative fallacy. Other methods can then be leveraged to understand what drives people and teams to make a certain decision or take a particular action. 

Typically understanding people’s motivation will surface root causes of problems. A common scenario involves conflicting drives becoming an increasingly complex problem. 

6) Understand what drives people to escalate problems

In the previous step, our focus was to understand the human drivers that caused the problem. In this step, we want to understand the drivers themselves that escalated into a problem. Some problems start as simple (or complicated), but over time, due to human interaction, they become complex problems.

These problems can’t be resolved without identifying the human interaction that caused them to escalate into a complex state. In this step, the focus is on personalities, but the previously mentioned tools should be used as well. For example, different personalities tend to turn a simple technical debate into a complex interpersonal conflict that impacts almost every aspect of a company. 

Understanding the causes of an escalation into a complex problem is a pre-requisite to resolution. If you don’t have this information, other simple problems might also become complex problems down the road.

7) Evaluate options by using simulation and select one

After identifying all the causes of a complex problem and identifying several options to resolve them, it is time to select the best solution. The challenge is that complex problems inherently have multiple parts and options so you can’t simply put them to pen and paper to find the right solution. Especially if it is not a generic problem but one you previously ‘solved’ at a systems level.

Some complex problem-solving methods (like Systems Thinking) provides the ability to use technology to capture all the parts, their connections, and the options for resolving a problem. Once captured in your database, this information can be modeled. Once modeled, you can run multiple simulations that will visually show the end results for each option. With a baseline model, you can run a test with current data to confirm the model against what has already taken place and simultaneously validate the results. You can then adjust the model to validate future recommendations from the insight and data you’ve gained. As you implement these recommendations, you can visually see what is happening while solving the problem.

Complexity means complex problem-solvers need simulations to vet the right solution rather than simply using pen and paper.

8) Implement the proposed option

This step is a straightforward way to resolve complex problems. As we have often heard, plan your plan and then work your plan. After the planning phase, decide on your priorities and which projects to implement and then do the leg work for implementation and execution. It is important to keep in mind a few principles when implementing change in a complex system:

• Changes created by a team or group will be adopted and implemented more easily than those imposed by an individual.
• Small changes over time are better than a big monolithic one. 
• The evolutionary process of small changes gives people time to adapt and eventually creates a solution that might not have been reflected in the planning phase but is actually a better fit for the environment.
• Randomness in the implementation process brings better results than following a linear implementation approach.

9) Set measurement system to monitor success and limit negative side-effects 

In conjunction with selecting the preferred solution, there should be defined metrics and other feedback loops to indicate if the proposed changes are getting the anticipated results. During (and after) the implementation step has been completed, metrics and feedback loops should be monitored to verify if the expected results are transpiring.

If you see increasing discrepancies between the expected results and actual results, there should be an effort to understand what is happening and how the baseline was modeled. Do not be afraid to pivot direction based on the new information or even revert to a previous state as a preferred option.

10) Monitor for recurrence of the same problem in different ways

Often, we seemingly resolve a problem only find it resurface in a different shape or form. Because we already know what to look for, an ongoing attempt to detect recurring symptoms should be put in place.

These symptoms help you realize that the same problem is about to re-occur. Early realization enables us to address and prevent reoccurring problems from happening in the first place.

Ignoring this step will result in unpleasant surprises. When dealing with complex systems we have limited ability to fully understand the impacts in the system. Only continuous monitoring of known symptoms can alleviate unpleasant surprises.