Fundamentals of Analysis – Pareto Principle

  •  
  •  
  •  
  •  
  •  
  •  

 

The 80/20 rule and how it can supercharge your improvement efforts

 

The Pareto principle, or 80/20 rule as it is called in business, is a fundamental idea behind many analysis efforts. The idea is enticing and simple; capture 80% of the potential benefit by fixing the most important 20% of the problems.

 

Pareto and power laws

 

Vilfredo Pareto came up with his namesake principle by gardening. He observed that when he grew peas, around 80% of the total peas came from 20% of the peapods. That idea stuck with him and he noted the same relationship in economic data. In the late 1800’s he showed that 80% of the land in Italy was owned by 20% of the population[1].

 

The relationship Pareto noticed is a power law distribution.  The Pareto principle, like other power laws, is scalable.  If you look at the group that owns 80% of the land, the top 20% of them will own 80% of the total land in that group.  That implies there is also a 64/4 rule where the top 4% of the people own 64% of the land, and a 50/0.8 rule, and so on.

 

Numerous studies confirmed his results over the 100+ years since his original work.  Similar power law relationships also exist in the natural world.  They are evident in disasters such as earthquakes and hurricanes.  You can also see them in the “tails” of ordinary characteristics such as height.

 

Note: If you have never heard of power laws and want to learn more, I recommend reading “The Signal and the Noise” by Nate Silver.

 

The 80/20 relationship is not a rule, but rather a guideline. If you looked at book sales for instance, you would see 97% of sales are made by the top 20% of authors[2]. The important thing is not the precise relationship, but rather that there is a concentration of benefits.

 

What about business today?

 

The 80/20 rule is common in business. 20% of your customers will be responsible for 80% of the good in your business, e.g., sales, profits, reviews. However, 20% will also be responsible for 80% of the bad, e.g., complaints, lawsuits, returns.

 

If the top 20% of your customers account for a much smaller amount of sales, that could signal an opportunity.  You should examine whether you can expand your relationship with that customer.  Those customers might like a premium product you are not currently selling.

 

When something goes wrong in your business, there are usually a few root causes for the majority of the problems. Finding and fixing those root causes can produce a dramatic change.  Good and actionable analysis often focuses on finding the most important root causes, aka the 20%.

 

How do we find the most important 20%?

 

Finding the most important root causes depends on the situation. The simple version is to summarize results by the true root cause. Yes, it is that simple.

 

When I say something is simple, I am not saying it is easy. The reason many organizations have not solved their 80/20 problems is that it is not easy to discover the true root cause.  That often requires rigorous analysis, asking hard questions, and tracking information over a long period of time.

 

Here is a hypothetical example. I have two small kids, and at least once per day one of them cries.  If you asked them why they are crying and tracked it for 3 months, you’d get the following responses.

 

Why are you crying? Frequency
She hit me 3
I dont like peas 2
I want to go to McDonalds 2
I dont want to go to school 4
I dont want to go to the store 2
He spit on me 4
He said his cat was softer than mine 1
I want a Captain America shirt 2
I want to wear Ninja Turtles pajamas 5
He wiped boogers on my seat 1
She said I wiped boogers on her seat, but I didn’t 1
She said My Little Pony was better than Pokemon 1
More silly reasons like these 124

 

Note: This is fake data. Except for the wiping boogers part, that happened.

 

That graph does not appear to offer much value. The reason is simple, it doesn’t. I used this example because the explanations that kids use are often comically bad. The reasons given in business sound better than these, but they may be equally useless.

 

Digging beyond the obvious with hypothesis driven problem solving

 

Luckily for me, it is obvious that my kids are not telling the whole story when I ask why they are crying. I start with the hypothesis that there are five underlying causes for their crying: they are over tired, they are hungry, Daddy is mad at them, Mommy is mad at them, or they are in physical discomfort.

 

As I am a father who wishes to remain sane, I seek to minimize the amount of time my kids cry. I hypothesize that the majority of the time they cry it is because they are overtired or hungry. I will seek to prove this and use it to justify a stricter meal and bed time schedule.

 

To prove my theory I will set up a measurement plan for capturing data on my kids’ reasons for crying. On a daily basis I will record variables such as their sleep and meal schedule. On a per cry basis I will record variables such as sources of pain, and my wife and I will assess each other’s perceived mood.

 

When I analyze the data I collected I find that my conclusion was not correct. We have an 80/20 situation, but hunger was not a part of it. In fact, they were almost three times more likely to be crying because I yelled at them than due to hunger.

 

Real reason for crying Frequency % of crying
Overtired 66 43.42%
Daddy mad at them 42 27.63%
Hungry 15 9.87%
Physical pain 13 8.55%
Mommy mad at them 4 2.63%
Did not want to leave 3 1.97%
Embarrassed 3 1.97%
Feelings hurt 3 1.97%
Frustrated 3 1.97%
Total 152

 

If I wanted to set up a plan to reduce their crying, what would I target? The evidence above tells me that I should set a proper sleep schedule and change the way I interact when I am angry at them.

 

Is there a better way to show this than a table?

 

Tables of data are often a bad idea in presentations. Listing out data can work, but leaders want the “so what” shown to them visually.

 

Luckily for us there is a chart designed for Pareto principle analysis, the Pareto chart. The Pareto chart is simple; it consists of an ordered bar chart paired with a cumulative percentage line chart.

 

There are instructions here on how to make a Pareto chart in Excel. They are good enough to help you make the chart, but I would not stop there. If you follow the directions, your chart will look like this:

 

 

There is nothing wrong with that chart, but it is not appealing.  I wish I could tell you that the design of your presentations does not matter. I wish I could tell you that your data would tell the story no matter what it looks like. But I can’t, the design matters.

 

Your presentation will convince more people if it looks nicer. Graphs that are easier to read get read more often. Graphs that highlight the most valuable points convince more people. If it was worth your time to perform the analysis, then it is worth your time to make the presentation look great.

 

Errors in a presentation are unforgivable. If your presentation has an obvious error your audience will discount or disregard your message.

 

Below is a quick attempt at a better Pareto chart for my imaginary crying analysis. Notice that I made the visual portion more prominent, highlighted the important columns by changing their color, and removed the confusing gridlines.  For this graph I preferred to show the column totals as well as the most important data point on the cumulative line.

 

 

Believe it or not, a consistent color scheme is helpful to make good looking presentations. The one I used below is the one I made up for The Analytics Dude. As you may have noticed from my website, I am a better analyst and writer than I am designer. Sorry about that.

 

Conclusion

 

The Pareto principle, or 80/20 rule, is a powerful concept that underlies many analyses.  The idea that you can get 80% of the potential value by only fixing 20% of the total problems is the basis for many process improvement efforts.  You should always look to see if an 80/20 solution exists.

 

Do you have a good story about the 80/20 rule? Do you have a suggestion for making the chart better looking?

 

Stand by for a case study where someone I mentored put the Pareto principle into action to save millions for his company!

 

 

 

[1] Pareto, Vilfredo; Page, Alfred N. (1971), Translation of Manuale di economia politica (“Manual of political economy”)

[2] Taleb, Nassim (2007), The Black Swan

1 thought on “Fundamentals of Analysis – Pareto Principle

  1. I don’t know whether it’s just me or if perhaps everybody else encountering problems with your website. It seems like some of the text within your posts are running off the screen. Can someone else please provide feedback and let me know if this is happening to them as well? This might be a problem with my web browser because I’ve had this happen previously. Appreciate it

Comments are closed.