Have you ever built a beautiful dashboard only to find out that you don’t know how to drill down to find any insights from the viz that you have just created?

Implementing the 5 why’s can help you solve your problem.

The 5 Why’s

The 5 why’s technique was developed by Sakichi Toyoda and focuses on exploring the cause and effect relationship for a particular event or issue. The goal of the technique is to find the root cause of the problem or event in question. The theory states that if you can ask and answer “why” 5 times, then the root cause of your problem or event will be revealed. Let’s start with a problem and apply the 5 why’s technique:

The technique is a general concept and will vary depending on the situation. It would be ideally used for basic and moderately complex problems.
For more complex problems, the 5 why’s alone might not reveal the root cause of the problem. In this case, you may need to ask “why?” a few more times or perhaps adopt a different analysis technique, eg. cause and effect analysis.

Using the 5 why’s technique

First, you will need to do is understand the problem at hand. If you have generated a few tableau charts, look for outliers!. Look for things that deviate from the norm.
Have a look at the chart below (using sample superstore data):

What strikes you as odd in the above chart? What doesn’t fit the pattern you would expect or that you see from the other data points?

We can see that although there is a product that has exceptionally high sales and profit, it would not be my biggest concern. There should be a strong positive correlation between sales and profit.
What is concerning though, are the three biggest profit losing items. All three items are all from the machine category.

FirstWHY?“. Why are our three biggest profit losing items all coming from the machine category?

Lets add Order ID to the rows shelf.
All three of our least profitable items have have much higher loss transactions than profit transactions.

Second “WHY?”. Why do we have such high losses for the majority of our transactions?

Now lets add sales to the marks card.
Highest Sales does not result in Highest Profits.

Third “WHY?”. Why are Sales and Profits not positively correlated?

Similar to the previous step, we are going to add discounts to the marks card.
Interestingly, the discounts being applied are the same, but the Sales value are very different.

Fourth “WHY?”. Why are the discounts the same, but Sales different?

Lets add Order date to the rows shelf.
We see that there is a difference in Year of Purchase. It would seem unlikely that in just one year, the price would increase so much.
Further, we will add Region to the rows shelf, and we will expand until we get to City where we see a difference.

Finally, the Fifth “WHY?”. Why does the City of purchase cause our profits to vary?

Our Findings

We will add one final dimension to the rows shelf, Ship Mode.
We can see that for the standard shipping mode for the products in question, we lose more money than first or second class shipping. Now there is another “WHY?” question here, but it would most likely require more specialized knowledge of the business to answer.


In conclusion, we can theorize that each city has its own pricing structure and consequently, we lose more money on some transactions compared to identical transactions in other cities. One idea to remedy the high loss Problem in the Machines Sub-Category is to set a standard pricing structure for all products across all cities/stores.

I hope you have enjoyed the post, and I will see you next time.

The Data School
Author: The Data School