The stars lined up this week leading to June’s short and sweet blog post: I started a new consulting project with Apple Inc., noticed a viz I was not familiar with in the news (about same client), and hours later Storytelling With Data published the monthly visualization challenge – leading to a great makeover opportunity.
 
A WSJ article used this chart (anyone recognize it?) to show the breakdown and change in Apple’s Services revenues:
​

My makeover goal was to highlight the key message more boldly - App Store and Apple Music sales were the top growth categories while iTunes sales declined most.

​Using a slope graph helped improve alignment and legibility, especially as it pertains to the categories with smaller contribution percentages (like Apple Music, and Apple Pay). Also, using color more sparingly helped pop the categories with the highest growth and decline over time.

There may be other chart types that would achieve the same result I was going after but the slope graph certainly fit the bill in more than one way this time!

​Which of the two visuals worked best in your opinion? Would you use another chart type to tell this story?

Waterfall charts are great ways to visualize pricing data, especially when used for internal decision-making in business/product groups and in guiding the Sales organization. Here’s why:

1. They start with at least one known data point the audience can easily anchor to: the product purchase price.
2. The chart is read from left to right, making it familiar and easy-to-digest for the audience:
   1. The first bar (far left of the x-axis) plots the product purchase price.
   2. The last bar (far right of the same axis) plots the either the net price or gross margin, depending on the objective of your analysis.
   3. The bars between the start and end points go up or down, depending on whether they increase or reduce the price (or profit) realization.

 
These visual cues flow and help the audience get your data story out quickly and effectively.
 
The waterfall chart below is an anonymized version I created for a Storytelling With Data challenge. It is based off of one I presented to inform executives and cross-functional teams about the magnitude and impact of pricing deductions offered by the Sales and Marketing organizations.  

What visuals have you used or seen used by others to represent pricing data?

A recent data makeover challenge, posted by Cole Knaflic of Storytelling with Data, compelled me to participate and share my data visualization online because it highlighted the intent of good data storytelling - to have an impact, be useful, and change the behavior of the audience in some way.

Participants were asked to think critically about an article and graph posted by The Economist about hurricane frequency in the US, and then propose their own visual of the same data. The original graph shows all 5 category storms, in a stacked bar chart, plotted over time since 1851. A trendline suggests a steep decrease in hurricane frequency, though the underlying stacked bar chart doesn’t necessarily corroborate the conclusion that hurricanes in America have become less frequent (it’s just too hard to tell).

Looking at their graph, two things stood out to me: the inconsistent year groupings on the time axis and the use of stacked bar charts to show trends per category. The time axis had years in groups of 10, except the last group which had only 7 and this led me to question why years were grouped at all.

My first inclination was to eliminate the year grouping, which caused the trendline to become less dramatic when looking at total hurricane frequency over time. The increase in major hurricanes (those with category 3 or above) also became less significant. The decrease trend in minor hurricanes (below category 3) was evident.

Regardless of the trend it still begs the proverbial question, “so what?” Are the readers, policy makers, researchers, homeowners, and others looking at this data equipped with information that is of use to them? Not really. 

Through further research, I learned that storm category can be misleading because it doesn’t capture the storm impact, or damage. Katrina, which was responsible for roughly $161B in damages, scored only 3 on the Saffir-Simpson hurricane category scale. The plot thickened when I plotted the storm damage cost data. At first glance it appeared storm damage costs increased over time, particularly since the 1990s:

​But then I plotted the normalized damage costs, a measure that neutralizes the impact of conditions like inflation, wealth, population, and housing units on the nominal damage number, the data show no increase in damage cost over time:

This story is compelling. Here;s the bottom line: 
- Overall there has been little if any change in the total number of hurricanes over the past 160 years.
- Hurricane damage on the other hand, increased dramatically.
- Why? Population increase in coastal areas where hurricanes hit is certainly one factor.

Now this is a story, and one that readers and stakeholders can do something with. Can we prevent or slow down population growth in high-risk areas? Can we put in place incentives to move or change insurance plans to reflect the potential damages? These are larger questions that have social, political, administrative and other implications. And they call to us as the data story unfolds.

On a personal note, this exercise reinforced my love of data analysis and data storytelling. This stuff is fun! It highlighted the difference between spewing out numbers and finding meaning in them. What are your thoughts? Would you share with me an example of when data reported to you lacked insight, or the opposite - when a good data story was told?