Pareto Analysis is a way of looking for the most common contributing causes to a situation. Using a Pareto chart to perform graphical analysis on your data can help you identify the biggest drivers to your process and appropriately prioritize your actions.
Analysis with the Pareto Principle
Often called the 80-20 rule,the Pareto Principle is a common ‘rule of thumb’ that 80% of the effects of something can be attributed to 20% of the drivers.
Pareto Analysis Principle Example:
- Profits – Many businesses discover that 80% of their profits are driven by 20% of their products. Thus it makes sense for them to focus on that 20% of those customers because that gives them the best chance to drive profits.
- Errors -Sometimes you can see that one aspect of your process is responsible for delivering 80% of your errors. If you fix that one process, you can achieve outsized results.
What if your data doesn’t show an 80-20 split?
Don’t worry. This happens all of the time in the real world.
Analyzing Multiple Groups in a Pareto Chart
Sometimes it makes sense to sum a few main items make 80%. This is a good strategy if you have one or 2 main drivers like so:
If you get 2 or 3 main drivers, then you can decide if you want to work on all of them or just one. You might decide that for your current project it makes sense to work just one and lock in the savings with that improvement. Then you can attack the others after you have that win. Reducing your project’s scope is often a good strategy for when you have a short timeline.
Other times it might be very expensive to fix the root cause of an issue. In that case, I’d recommend creating a cost benefit analysis to see if your project can generate the savings necessary to make doing the fix worth it. If it is, great! If not, you might see if you can either change your process to avoid the situation or focus on the other items.
What if the Pareto Chart Is Flat? (Or, what if nothing pops on the chart?)
Sometimes the chart doesn’t show one item or even a few items as outsized drivers in a process. Sometimes there is very little to distinguish one group from another.
This usually indicates that the variable you are investigating isn’t the right driver for the relationship. Try different categorizations or a different data attribute in your analysis.
What to do when the ‘Other’ or ‘NA’ category is the biggest
This is a huge pet-peeve of mine. Sometimes people will not label their data correctly or at all. Then, when you make a Pareto chart, the other category or NA is the biggest one on the chart.
This reflects a lack of data. The only course of action is to go back and correctly label your data.
Pareto Analysis Example: 80% of defects as other.
Let’s say that a software development agency was mapping out the root causes of bugs in their software. Some bugs were found to be errors in the data, others in the user interface code, and still others were missed requirements. But in their error tracking software people did not select the right root cause and the system defaulted to ‘other.’
When the team tries to do process improvement, they are faced with a graph that looks like this:
Sure, the 3 categories are represented, but that ‘other’ category dwarfs them.
It makes no sense to ignore the ‘other’ category. It’s just too big.
The only appropriate thing to to is to go through each of the tracked bugs and update the status to the appropriate category.
Have you performed Pareto analysis? What patterns have you seen?