Data visualization is an important part of analytics. Analytics effectiveness and impact depends on visualization skills of two kinds: the ability to create visuals and ability to understand visuals. The real value of visualization does not come from creating visuals, but from understanding what they can tell you. Making them understandable, of course, depends on the abilities of those who make visuals. But value and impact are derived from those who read them, interpret them, and use them to drive conversation, collaboration, and innovation.
With the language, we learn reading and writing as separate but related skills. Similarly, with visual language we need to learn understanding (reading) and creating (writing) as distinct but related skills. There are many books, courses, and other resources that teach people how to develop data visualizations; but few that teach how to read and understand them. My goal in this brief piece is to describe what I believe to be the principles of finding the right meaning and the important insights in data visualizations. To add interest and a bit of challenge, my goal is to describe those principles using only words and no visuals.
Most frequently, readers eyes go right to the center of a graph – to the colored lines, bars, or bubbles – then begin to try to make sense of the shapes and colors. There is a better and more systematic way to find meaning in charts and graphs.
Start with the Title – Every chart and graph should have a title. If the title is missing then you have no context and are almost assured that you’ll struggle to understand the meaning. Question the title: is it clear and non-ambiguous? Does it give you a sense of the purpose of the graph? Does it provide enough information to truly frame the message? Consider, for example, these chart titles.
- Fiscal Year End Sales
- Cumulative Sales at Fiscal Year End
- Cumulative Sales at Fiscal Year End: Same Store Sales FY 2014 and 2015
Clearly the third title provides the greatest amount of information to help in understanding the graph. You have a lot of information here even though I have not actually presented a graph. As a reader of visuals, don’t just accept what the title tells you; ask what it doesn’t tell you.
Understand the Axes – Most charts and graphs work on a coordinate system with horizontal (x) and vertical (y) axes. Look at each axis independently first. Is the axis labeled? Does the label make sense? What business concept does it illustrate? Does it represent quantitative, categorical, or time-series variables? What quantities, time periods, or categories are used?
With independent understanding of each axis, next look at the two axes together. What relationship between two business concepts is illustrated? Does the relationship make sense and provide useful business information? Relating sales amounts to time periods, for example, is a sensible association of business concepts.
Examine the Scales – Now consider the scale for each axis. This is an essential step to make sense of the numbers represented in the graph.
If it is a quantitative scale, what are the units of measure? Does the axis start at zero or at another value? Are only positive values shown, or also negatives? What do negative numbers mean? Is zero non-arbitrary (a ratio scale such as dollars where zero dollars is absolute) or arbitrary (an interval scale such as degrees Fahrenheit where zero degrees is dependent on the measurement method)? Is the scale linear or logarithmic?
If it is a time scale, what time span is represented? What are the intervals between units of time? Are the intervals uniform, graduated, or ragged?
If the axis illustrates categorical variables, look at it a bit differently. Categories aren’t really a scale, but you still need to question the labels on the axis. Are the categories all-inclusive or a subset of all possible categories? Are they mutually exclusive or overlapping?
Look for a Legend or Key – Does the graph contain a legend that explains the meanings of shapes, colors, and patterns? When a graph contains both rectangles and circles to represent data points, what meaning is encoded in each shape? Similarly, what is the difference between solid lines and dashed lines? Or between red bars and blue bars? Seek explanation of the colored lines, bars, and bubbles before looking at the core of the graph.
Look for Data Sources – Does the graph tell you what data was used and where it was acquired? It is a common practice to place data source information at the bottom of a graph, but it may also be found in the legend or embedded in the title. Or it may be missing entirely. Consider data sources to make your own judgment about the accuracy, truthfulness, and reliability of the data.
Look at the Data – The graph encodes data as points, lines, bars, columns, bubbles, etc. These shapes are the data in the data visualization. What does the visual show you about the data: comparisons, proportions, relationships, whole-part assemblies, locations, movement, patterns, or trends? Thoughtful examination of the data – with context of title, understanding of axes and scales, descriptions of encodings provided by a legend, and level of confidence in the data – will guide you to find real meaning and interesting stories in data visualizations.
