Effective Data Visualization – The Eradication of Ruler Reports
I once had an efficiency consultant tell me that he went into an office looking for a ruler left out on someone’s desk. This was a sure sign that person received one or more reports that were so large and so complex, it was necessary to lay a ruler across the report in order to read it properly. These “ruler reports” were the first candidates to be redesigned and replaced.
Ruler reports do not lead to effective decision making. In fact, the confusion they create can lead to poor or even incorrect decisions. Equally damaging, ruler reports slow down the decision making process because of time dedicated to deciphering the content. Ruler reports are not effective data visualizations.
The following guidelines for creating effective data visualizations will help you eradicate ruler reports in your organization.
Effective Data Visualization Guidelines
Effective data visualizations allow viewers to quickly and accurately analyze data to make better decisions. Effective data visualizations follow three guidelines:
Simple visualizations are clean and uncluttered. They present only the necessary information for the decision making they are intended to facilitate. It is also helpful if they are pleasing to the eye. Of course, this last requirement is to some extent a matter of taste, so is open to interpretation.
Interactive visualizations provide the viewer with an opportunity to easily explore data and gain additional detail, when required. This can include everything from implementing a mouse-over tooltip to providing parameters and slicers allowing the viewer to change report content. It may also include the ability to drill down within a report or to drill through to a different, more detailed report with a single mouse click.
Self-Interpreting visualizations provide visual cues that give meaning to the data being presented. This can include something as simple as a line denoting a goal value or an indicator showing performance against budget, which can be extremely helpful. Trend indicators and trend lines provide context that can be very useful when drawing conclusions from a set of data.
An Example of Effective Data Visualization
To make these concepts more concrete, let’s look at an example. Figure 1 is a Manufacturing Report showing the number of products that were produced and accepted by the quality assurance (QA) process, the number of products produced and
rejected by QA, and the number of products rejected as a percent of the total number of products produced. With all of that data, for an entire year, this is certainly a ruler report.
Not only is the report cluttered, making it hard to read, but the headings are difficult to interpret. The products are denoted by product number alone with no description. In addition, the column headings use a concatenated “YYYYMM” format, making even the month tough to decipher. Even the title is overly generic.
Effective Alternative #1
Now that we’ve identified some of the issues with the report in Figure 1, let’s take a look at an alternative created with the effective data visualization guidelines in mind. This is shown in Figure 2. Prior to creating the report, the author pressed the user to determine what information is primary to the decision-making process supported by this report. The feedback? Analysis is based almost exclusively on the percent rejected with
the actual number of products produced, and actual number of products rejected as secondary pieces of information that were taken into account less frequently.
With this in mind, the report was simplified to display only the percent rejected. The actual number of products produced and the actual number of products rejected are available through a mouse-over tooltip. In addition, the number of decimal places being displayed for the percentages has been reduced to two digits to the right of the decimal. This is all that is necessary for the required analysis.
While still including a fair amount of numeric data, the layout in Figure 2 is much simpler and easier to understand than the report in Figure 1.
A number of interactive items have been included for the report in Figure 2 to increase its effectiveness. It was determined that different managers are responsible for analyzing the percent rejected for different product types. Consequently, a product type parameter has been added so the report shows data for only one product family at a time.
It was also determined that the data can be initially analyzed at the product subtype level with occasional investigation at the product level. Therefore, drill down is implemented from product subtype to product. Finally, as noted earlier, a mouse-over tooltip has been added to display detail information.
The report in Figure 2 contains a number of features to assist with interpretation of the data. First, the month designations in the column headings are presented in a format that is much easier to read. In addition, the product name is included along with the product number in each row heading. The report title has been changed to better reflect the data contained in the report and the goal for the metric contained in the report is included in the report header area.
Finally, a color has been added to the cell background to immediately signify the value relationship to the stated goal. The farther the value is beyond the goal, the darker the red color. You should keep in mind when using color to assist with data visualization that some of your report readers may be color blind. Use the intensity of a single color rather than multiple colors to express meaning. This has the added benefit of translating well to black and white printing.
Effective Alternative #2
Another approach to creating a more effective alternative to the example report is shown in Figure 3. This alternative stems from another user’s statement that the example manufacturing report is used mainly for finding reject percent outliers in the current month’s data and negative trends in reject percent in the data leading up to the current month. This type of analysis is best facilitated by a dashboard with drill through
capabilities. Figure 3 shows the Percent Rejected dashboard along with two levels of drill through reports.
When done correctly, a dashboard is the ultimate in simplicity. It includes a value indicator showing the state of a value. It may also, optionally, contain a trend indicator. In the dashboard shown in the upper left in Figure 3, the large indicator symbol shows the value of the percent rejected relative to the goal. Note that both the shape and the color of the indicator convey the message here so the dashboard can be easily interpreted by someone who is colorblind (Granted, a symbol key or some training may be required to learn the meaning of the symbols absent any color cues). Arrows indicate whether the value is trending up, down or remaining steady. Again, both arrow orientation and arrow color convey the trend status.
Drill through from the dashboard to a summary report and from the summary report to a detail report is shown in Figure 3. Users can pick out a product type on the dashboard that is causing concern and click it to view information on the constituent product subtypes. In the same manner, users can pick out a product subtype that is causing concern from the column chart and click it to view detail information on the individual products within the product subtype. Mouse-over tooltips are also implemented for the dashboard and the product subtype report to display more detail.
As discussed previously, the dashboard uses indicator shape, color and orientation to provide instant interpretation of the status of a value or trend. The product subtype column chart includes a line showing the goal for percent rejected. It also includes shading, along with emoticons, to show which side of the goal line gets a positive interpretation and which side gets a negative interpretation.
The detail report includes several indicators to assist with interpretation. The percent rejected value column includes a bullet chart to show relative value as well as an indicator to show the state of the value relative to the goal. Finally, it includes a spark line to show the trend of the value over the past 12 months.
Effective Data Visualization Is Easy and Fun
As the example demonstrates, it is not difficult to create effective data visualizations. Just remember the basic guidelines. Keep your visualizations:
- Interactive, and
Besides leading to quicker and more accurate analysis resulting in better decision making, the visualizations in Figures 2 and 3 are more fun to create and much more fun to work with!
So follow the guidelines, be creative, have fun and visualize better!