This month’s question
What are the different tools used to conduct root cause analysis (RCA)? How do they work?
Several tools can be used to conduct RCA. They range from simple, easy-to-use tools such as why-why analysis to more complex tools such as design of experiments. A simple tool can be just as effective as a complex tool when applied effectively at the right problem-solving phase.
At a high level, RCA involves defining the problem, analyzing the problem and developing improvement actions to prevent recurrence. An extensive amount of information about quality tools is widely available through textbooks, publications, papers and presentations, but here are some of the most widely used tools and applications, categorized by problem-solving phase.
Phase: Define the problem
Problem solving starts with an effective problem definition. Without an effective problem definition, valuable resources will be wasted by the problem owner, who won’t be able to resolve the issues in the timeframe he or she proposed to the organization’s management.
To effectively define the problem, use is/is not analysis to understand the what, where, when, who and how of the problem:
- What happened? What didn’t happen? Knowing what happened prevents resources from being wasted on investigating nonissues.
- Where did the issue happen? Where didn’t the issue happen? Knowing this information concentrates your efforts on a specific location, process step or activity in a project work breakdown structure.
- When did the issue happen? When didn’t the issue happen? This information accelerates the investigation efforts and determines whether there were any contributing factors around the timing of the event. The most common example is unintended effects from a previous change rollout.
- Who is affected (internal and external stakeholders)? Who isn’t affected? Knowing who’s involved solicits key information for the investigation and targeted communication to those who were affected.
- How many issues are there? What impact did they have? What impact didn’t they have? This information helps you determine whether you must involve certain stakeholders, such as management, customers, regulators or public relations. It also helps you plan the appropriate resources and corrective actions based on the magnitude of the impact.
There are many other supporting tools used to provide objective data in a problem statement, including check sheets, histograms, Pareto charts, defect concentration diagrams, measles charts, timeline analyses, value stream analyses and flowcharts.
Phase: Analyze the problem
Brainstorming is used to generate many ideas and inputs. Although it can be used at all stages of problem solving, it is used more extensively when identifying root causes and improvement actions. Brainstorming generates many causes relevant to the effect of the problem. The purpose of brainstorming is to generate many inputs through creative thinking. The useful ideas are narrowed down at a subsequent stage.
A check sheet is a structured tool that captures and analyzes data. In special cases, the check sheet is designed with a certain level of segmentation (rows and columns) to visualize data distribution through tally marks. Due to its simplicity, it is a widely used tool, and application at the point of use is possible with minimal resources and little training.
Concentration diagrams and measles charts are used to focus the analysis and improvement efforts where the defect occurred most. Examples include the construction site with the most safety incidents or the city location with the highest frequency of crime.
A cause and effect diagram (or fishbone diagram) is the most widely used RCA tool. It helps you visualize the inputs collected during brainstorming by categorizing them under headings. The most common headings are people, process, material, equipment, measurement and environment. A cause and effect diagram also can be applied to nonmanufacturing applications where there is no physical material, equipment or operating environment. Service sectors, for example, use the headings information, infrastructure and organizational culture.
A histogram identifies the central tendency and spread of a process. Shifts and changes in the variability of a process are easily understood without advanced techniques. Another advantage of a histogram is the ability to visualize the presence of an abnormality in the process locating multi-modes and skewness of the distribution.
A Pareto chart helps you visualize the 20% of issues causing 80% of the overall effect. Coupled with severity weighting, it can be even more useful to focus on the issues with significant effect and not just those with the highest occurrence frequency.
A scatter diagram analyzes the relationship between two variables—an independent variable (X axis) and a dependent variable (Y axis). The plot provides an easy-to-understand visualization of the nature and degree of correlation between the variables, which is important to RCA. Users should be aware of causation and correlation before interpreting the results of a scatter diagram.
Stratification is a useful but underused tool. It provides a visual display of distribution of data stratified by key process variables, making it easier to spot abnormalities and focus improvement efforts. To perform this analysis, metadata must be attached to the process data collection. This tool also can be used to quarantine affected outputs quickly from the time incidents are reported by key stakeholders.
Phase: Develop improvement actions
When developing improvement actions, use the five Ws and two Hs (also called 5W2H) approach to determine the who, what, where, when, why, how and how much of the actions:
- Who will do it? Who is responsible for taking action? Who will independently verify the effectiveness?
- What actions will be taken?
- Where will it be done? (Location or area.)
- When will it be done? (Timeline and dependencies.)
- Why is it being done? (Justification, reason and link to root causes.)
- How will it be done? (Method or process.)
- How much will it cost to take the actions? How much will the organization benefit in the long run?
A Pareto chart also can be useful for visually displaying before-and-after data (improvement) by presenting the information side by side. Similarly, histograms, defect concentration diagrams and various other statistical tools can be used to demonstrate improvement and changes.
A control chart is used to monitor the statistical stability and variability of the improved process. When statistical rules are applied to this chart, it serves as an effective tool for proactively monitoring for out-of-control situations before the process starts to produce nonconforming outputs.
It’s important to note that the application of these tools is not the ultimate goal in problem solving. Rather, it is the resolution of the problem and preventing recurrence. Knowing which tool to apply and how to apply it helps achieve that.
ASQ, “Quality Tool A to Z,” https://asq.org/quality-resources/quality-tools.
Ramu, Govind, The Certified Six Sigma Yellow Belt Handbook, ASQ Quality Press, 2016.
Tague, Nancy R., The Quality Tool Box, second edition, ASQ Quality Press, 2010.
This response was written by Govind Ramu, program manager, End-to-End Closed Loop Learning, Google, Mountain View, CA.