Cheese and Onion – Frameworks for Analyzing Project Failure

When it comes to improving success rates, the commercial aviation sector has been one of the most successful. In 1930 taking a commercial flight was risky business. When you board a flight today, the chances of an accident taking your life is about 1 in 30 million. Through rigorous accident investigations and a willingness to challenge every facet of the problem, the industry has continually improved its safety record.

Sadly, success rates for the projects in today’s businesses are far worse and evidence suggests that projects are equally prone to failure today as they were twenty years ago. Chances are many reading this will have experienced a troubled project at some point in their career and some may be in the middle of such a situation right now!

One of the reasons the aviation sector has been so successful is because they have adopted a framework for investigating the causality behind their failures. That framework recognizes that the origin of failure can be attributed to either personnel or systems factors. The personnel perspective attributes failures to the errors, omissions or mistakes of the person whose actions directly resulted in the failure (often the pilot, the air traffic controller or the mechanics). The solution to such failures is to increase training or to ensure only qualified resources are assigned to perform the tasks. The systems perspective looks at failure in terms of the context within which the failure was able to occur. As such it focuses on policies, procedures and other organizational elements to identify the root causes that established the context within which the mistake was able to happen. The solution to such failures is to modify the processes, procedures or factors that allowed the mistakes to occur.

Figure 1 - The Swiss Cheese model (click to enlarge)

Figure 1 – The Swiss Cheese Model
(click for larger version)

A common framework used for such analysis is James Reason’s “Swiss cheese” model (Figure 1). Reason’s model says that when investigating failures we should look for causes at four separate levels: Organizational influences, unsafe supervision, preconditions for unsafe acts, and the unsafe acts themselves.

Applying that to a project context we may find that a team member made a mistake when defining the scope of the project (an unsafe act), however the organization’s process for identifying stakeholders and gathering requirements was ineffective (a precondition for the unsafe act). Peeling back the layers we may also find that the project lacked a Sponsor with the skills to oversee the project (unsafe supervision) and beyond that, departments acted as “silos” thereby hindering communications needed to ensure the full extent of the project’s scope was understood (organizational influences).

To prevent failures the model says that organizations should put in place a set of defences (processes, procedures, checks and balances, etc) at each level. In the model, these defences are represented by slices of cheese. The levels of defence work together to form a system that is designed to prevent failures and catch mistakes before a problem can occur.

Because no system is perfect there are weaknesses in each level. These weaknesses represent opportunities for mistakes to be made. In the model the weaknesses are thought of as holes in the cheese (hence the name “Swiss cheese”). Reason says that the system of defences fails when “a trajectory of accident opportunity occurs” (i.e. failure occurs when a mistake is able to pass through all four levels of defence undetected and uncorrected).

When looking at project failures in today’s businesses it is rare to find such a systematic analysis of cause and effect. When faced with a project failure some organizations simply ignore the issue and pretend it never happened. Some fire the executive in charge (I have strong anecdotal evidence suggesting project failures are the leading cause of senior executives being forced out of an organization) and others gloss over the issues assuming the next project will go better. All of those approaches are a set up for the next failure and lack the rigour of an effective management approach.

If organizations are serious about improving their success rates they need to take a multi-layered approach. Rather than blaming “the sharp end of the stick” (the individual whose actions resulted in failure) they need to have a culture that says that quality work and the prevention of failure is important (layer 1 in the defences), they need to put in place effective project governance (level 2), they need to adopt processes, procedures and tools that help prevent mistakes being made and help identify them as soon as they are made (level 3) and they need to provide appropriate training to ensure workers can do their jobs properly (level 4).

Some people like to use the Swiss cheese model when analyzing failures, others simply use the analogy of peeling back layers in an onion. However you think about it, the key to success lies in going below the surface and addressing the deeper seated issues. The truly successful organizations are those who are able to see these deeper problems and take action to build up their defences.

Do such organizations exist? Absolutely. The following article “Intellectual Infrastructure” outlines the management practices used by one such organization. Although the article is about a software development organization, many of the lessons are universals that would have application in a wide range of different organizations.

Reference: Human error: models and management – British Medical Journal – Mar 2000 – J. Reason