Visual Discovery and Interactive Dashboards

By Rado Kotorov

Many people have asked about the difference between Visual Discovery dashboards and interactive dashboards. And rightfully so, since interactive dashboards have become a corporate must-have as a means to identify business problems and drill down to details. But the term is becoming more generic, and therefore losing some meaning, as Business Intelligence vendors claim to offer interactive dashboards. So it is important to point out and clearly define the differences and the value added by Visual Discovery dashboards.

Dashboard interactions fall in three categories:

1. Drill-down on coordinated/linked charts and reports
2. Visual querying of coordinated charts and reports
3. Visual "What if" analysis

The visual "What if" analysis is a special case of interaction. The end user is provided with sliders to manipulate (increase or decrease) select values, and see the immediate impact of the changes on the other categories and/or the total. As shown in Figures 1 and 2, manipulating the growth rates for any sales category is immediately reflected in the total sales gauge.

Since "What if" is a special case, the more important question is how visual querying and analysis is different from drill-down interaction.

In essence, drill-down takes the user to the details of a selected category. It is a two-step process:

1. Focus on a particular category of interest.
2. Click-through to the details on this category.

The click-through changes the content of the page and the graphs to display information only about the selected category. Let us take a look at an example. Figure 3 shows sales analyzed by two dimensions – gender and age – with each pie displaying the percentage of total sales for the dimensional categories.

Drilling down on "male" would re-scale both charts, as shown in Figure 4. The male category is rescaled to 100 percent, or it occupies the entire pie. The Age Group slices are rescaled to show only the percentage of male sales within each age group.

If users want to analyze the percent of sales to females within age groups, they would have to return to the original page, shown on Figure 3, and drill down on the female category. This two-step process does not allow for easy category comparisons, such as male/female comparison, or for assessing the relative importance of sub-categories (percent of total category), such as the percent of total sales for males ages 33 to 45.

The visual querying is a three-step process:

1. Category selection
2. Display update to show the relative importance of the category (percent of Total Category)
3. Drill-down, which is optional

Let me demonstrate the difference in the user experience and the analytic power that this extra step would make in the dashboard that we just discussed.

The initial dashboard will have the same look and feel as the dashboard shown in Figure 3. However, upon selection of the male category, the rest of the categories are not removed from the charts. Instead, the unselected categories are grayed, so that the user can see the relative volume of sales to males within the total category sales. In other words the user can easily see that 70 percent of the purchases in age group 33-45 are made by males. Upon a drill-down, the charts displayed on the visual analytics dashboard will become identical to those in Figure 4.

The intermediate step, which was just described, allows quick comparisons of categories in multidimensional data. Its power becomes even more obvious as the number of dimensions and categories within each dimension increases (see Figure 5). Combined with some robust chart linking, filtering, and coloring techniques, it not only expands significantly the number of variables that users can navigate, but also reduces the time required for the analysis.

Linking, filtering and coloring techniques will be the topic of a future article.