Creating accurate, honest and useful visualizations of data is one of the trickiest endeavors in business. Basic measures, such as revenue over time, are communicated nicely by the built-in charting capabilities of Excel, Numbers, Google Docs or any number of apps. But even then, temptation lurks in all the colors, styles and formatting options. It's all too easy to get carried away, adding superfluous labels, legends, bars, lines and even data itself.
As in marketing, product management, corporate strategy and even life, the best solution is to put yourself in the other person's shoes. Who is the audience, what's important to him and why does he need the information? For example, to an auto manufacturer, tracking monthly unit sales is critical, but arguably more important at a strategic level is their share of the total market. Increasing unit sales are great, but if the market is growing at a faster pace than sales, their market share is declining, which is information that must be communicated.
After establishing the who, what and why, then comes how you communicate the data, which is where charts come in. There are myriad options for presenting data to the team managing our example auto manufacturer, but the choice should be based on ensuring the audience receives exactly the data they need as simply and understandably as possible. Unit sales and market share, for example, are immediately grasped and understood when charted with two axes, much less so if just in a spreadsheet.
I recently ran into this in an analysis I was writing for my company. We conducted a survey of residential investors to better understand their recent experiences in the market, as well as their expectations for 2014. One question was, "In 2013, how many investment properties did you buy," followed by, "In 2014, how many investment properties do you plan to buy?"
A straightforward and standard Excel-type solution is a column chart that compares 2013 to 2014:
The chart above accurately illustrates our survey respondents' optimism about their property acquisitions this year, with only 5% expecting to exit 2014 having not bought any investment houses. However, the amount of data contained in this chart is slight. If you're in the business of writing loans to these investors (as we are), it's encouraging but not actually helpful to know that people expect to buy more. More helpful would be to know exactly who expects to buy more, and how much more. For example, how many of the people who bought 1-3 last year expect to buy 4-6 this year, versus another 1-3?
To inject more meaning into the chart, I sought to show where members of each 2013 group migrated in 2014. My first attempt was a stacked column chart:
Answering the example question above, 18.75% of those who bought 1-3 properties last year expect to buy 4-6 in 2014, represented by the yellow section of the second column. So this chart has the desired information, but it's confusing. Charts are like jokes: If you have to explain it, it doesn't work.
The other problem with this chart is that I lost the 2013 and 2014 percentages for each group. What the reader of this chart would most appreciate is knowing those percentages, as well as generally understanding the migration between groups. The actual numbers behind the migration are overkill for the intended audience.
Therefore, I went old-school and sketched out a chart that would show everything I wanted:
Nothing fancy, just a quick visual check to ensure that what was in my head could conceivably result in what I wanted. I then worked out the tick marks I would need and cranked up Illustrator:
Now we have the percentages for each grouping, as well as a sense of the migration between those groups. We can quickly see that the bulk of investors expecting to buy 4-6 properties this year bought 1-3 last year, equipping us with knowledge to tweak our products and marketing accordingly.
Is the chart perfect? Nothing ever is and even looking at it now, a list is forming in my head of things I'd modify and enhance. But it capably visualizes and conveys the needed data for the intended audience, which is all we can ask of a chart.