In this article we will explain how to choose the right chart for your data. Or more precisely, for your message.
There are 3 fundamental rules that you should always follow when deciding which chart to use. If you learn these 3 rules, there is a good chance that you will get it right no matter what the context of your data is.
So, without further ado, let us start with rule no. 1.
#1 Display time horizontally, category items vertically
Let us focus on most frequent business charts, such as column and bar charts, line and area charts, dot plots or pin charts, plus-minus (variance) charts, etc. Your first decision should always be whether to turn them horizontally or vertically.
Here's how you should make your decision:
Of course you can visualize other types of data categories with horizontal charts. But you will probably run into problems with labelling, category spacing, etc. If you really want to make an exception to this rule, do it only when the number of data categories is very small and their labels (names) are very short.
1. It is easier to display category labels
2. You can insert them into tables thus creating combined table-chart reports
Charts can also have circular (polar) orientation, such as pie charts or "spider" (radar) charts. Additionally, we also have 2- or even 3-dimensional charts orthogonally oriented in two-dimensional space, for example scatter plots (XY charts) or bubble charts.
Here is a summary of all possible chart orientations and when to use them:
|Orientation||When do I use it?|
|1.||Horizontal||Use only when your data represents time series (years, quarters, months, weeks, days or date/time values)|
|2.||Vertical||Use if you have discrete items (categories or structure data, such as products, customers, accounts, etc.)|
|3.||Circular||Avoid pie charts whenever possible. Use only when the number of categories is very small (2 or 3 categories)|
|4.||2D||Use when you have 2 or 3 variables in your data, such as in bubble charts or scatter plots|
#2 Choose the right chart for your message
BI/reporting/visualization tools usually offer a large set of chart types to choose from. So it is of utmost importance that you understand and predict the nature of the analytical task that your readers will perform when observing your charts. You should always ask yourself: What message am I trying to convey?
For example, you might simply want to show an interesting trend over time. Or enable your readers to compare sales values of different products. Yet another time your readers will need to understand the variance to budget, growth from previous year values, contribution of specific products' variance to the total variance, etc.
Here is our cheatsheet that will help you select the right chart for most common tasks that business users typically perform:
#3 Determine the optimal chart shape for your data
Once you have chosen the correct chart orientation and the intended task of your visualization, you are ready for your last major decision: Which chart shape should I use?
Instead of always just using the identically looking "default" column chart regardless of what it actually represents, you should use different chart shapes for different types of values (KPIs). This is called shape coding:
For example, the columns in the Price chart are much thinner than the ones in the Revenue chart. This is quite intuitive - revenue values are usually way bigger than prices since price is actually a ratio of revenue, that is, it is divided by sales quantity (a value/volume division). Similarly you can use "pin" charts (yes, the "lollypops" 🙂 ) for percentages since they are value/value or volume/volume divisions and are thus without any unit (just indexes, %). So it seems intuitive to use chart shapes without width for indexes in business reporting. Once readers understand this, they will instantly know what they are looking at, even without reading the labels. Especially in a repeating process such as monthly reporting, this can come in very handy.
Similarly you can use patterns to code data scenarios like Actual, Budget, Forecast and Previous Year. For example, here is how budget values can look like:
Check out IBCS (International Business Communication Standards) for more great ideas like this.
Achieving consistency in business reporting is critical!
It is relatively easy to select the perfect chart shape when you only have one chart. That's not rocket science and many researchers and authors have written guidelines on how to do this. But business reporting is so much more than just designing one single chart. In fact, most times you will have to use several charts at once to get the full message across to your readers.
Often you'll need to design a report with many pages, a BI dashboard with several screens, etc. This is when consistency in your design becomes of utmost importance.
Just as traffic signs consist of consistent colors and shapes that always mean the same thing, elements in business reports should be consistently designed to depict or "code" the same meaning with always the same shapes, colors, patterns, etc.
Emphasize single values or trends?
The shape of a chart itself determines what we will see and how we will interpret the numbers. Several "gestalt" laws govern our visual perception. That's a fact. Different chart shapes will convey different messages to your audience.
For instance, some charts are more suitable for emphasizing single values, while others are better at displaying trends. Let's illustrate this with the following examples:
While we can see the overall trend, the vertical tendency and clear indication of bar heights support us in comparing single values (compare heights between two single values, such as Jul to Aug or Nov to Dec).
Use this chart shape when the number of periods is small and only when the chart is big enough.
The flow of the line emphasizes the trend, while the dots mark the height of a single value.
There is much more emphasis on the continuation or the flow of the values (trend analysis), but still some support for single value comparisons.
There is no explicit graphical support for single value comparisons, making it very hard to estimate the differences between particular months.
What is left is the continuity of the line, clearly exposing the trend in the data.
So, the perception of different chart shapes is ordered from vertical bars ("columns") to more "horizontally" oriented lines:
Still puzzled which shapes to use? Here are two rules of a thumb:
- Use chart shapes on the left when single values are important (e.g. annual or monthly figures) and when you expect that users will compare values between each other (e.g. current month to previous month). Also note that this is only meaningful when charts are big enough that you can comfortably display the columns and labels.
- Use chart shapes on the right when you want the users to focus on overall trends. Also, choose line charts when your space is limited (e.g. small chart sizes, small category widths, etc.).
Check out Zebra BI to create the charts from this article in MS Excel with just one click.
Just remember, before you click to insert your perfect chart, think about the 3 fundamental rules!
Get the full whitepaper in PDF format - it's free
PDF format, 18 pages, Jan 2015