Designing an Effective Dashboard is Like Creating a Steven Spielberg Film
The success of a CPM (corporate performance management) dashboard is dependent on several key ingredients. In past columns, we discussed the importance of profiling the user, selecting the appropriate framework and leveraging icons and color harmony to provide a holistic experience. This month, the focus is on selecting the appropriate graphs for displaying the KPIs (key performance indicators) and data.
Graph selection and presentation are critical to the success of a CPM dashboard. Not only must the correct metrics be selected, but they must be displayed in a format that is easy to interpret. Not difficult you say - just throw together several gauges, bar charts and line graphs on a screen palette and you're ready to roll. Wrong! Follow this formula and your dashboard is doomed to certain failure. Designing an effective CPM dashboard is much more like creating a Steven Spielberg film. Extensive thought and preparation is required up front to ensure a big splash at the box office (or executive board meeting). No dashboard is successful without the right graphs. "One glance analysis" is the operational mantra of today's busy business executives, and comprehensive well-designed graphs enhance the deliverable.
The centerpiece of any dashboard design is the KPIs and how they are captured and combined visually in graphs to reflect the health of the business. When selecting an appropriate graph in the good old elementary school days, the choices were quite simple - the basic graph styles included a bar chart, a line graph or a pie chart. Now, the choices are far more diverse and complex, as the landscape has exploded to include alternative graph styles and types, multiple scales and dimension views, and alternative presentation options. To streamline the graph selection process, the graph styles have been categorized into four major areas.
Basic: Core graphs such as horizontal bars, column bars, pie, area and line graph types where both singular and multiple data series are compared. The list is considerably expanded when the many alternate flavors such as the clustered, stacked, paired and progressive variations are included.
Difference: Graphs where the differential between two data series, rather than the absolute value, is of paramount importance. This can be expressed either as a direct comparison between several data series (change range) or as a comparison to a reference baseline (deviation bar).
Distribution: Unique graphs that profile the spread of the data through frequency bar charts (histogram); metric descriptors such as median, mean and percentiles (box-and-whisker); and daily movement indicators of stock market activity (HLC - high-low-close).
Specialized: Collection of graphs that focus on specific business areas that include planning (bubble charts), competitive analysis (radar), financial profiles (mosaic chart) and three-dimensional plotting (trilinear).
The graph selection road map in Figure 1 (see page 69) provides a framework that illustrates many of the potential permutations available for the four major graph styles. Keep in mind that this list is a guide rather than a comprehensive list of all options; for instance, "specialized graphs" can easily be expanded to include more business-function focused and industry-specific examples.
In selecting (or designing) graphs for the CPM dashboard, it is important to choose the appropriate graph style to meet the specific needs of the business application. This requires that several key criteria such as focus of the analysis, time continuum, number of data series and the data metrics all be incorporated in the graph selection process.
Focus of Analysis. How the data will be analyzed is an important criterion for identifying the appropriate graph type. Data series can be compared to each other (e.g., cluster bar, paired bar or change bar) or to a reference baseline (e.g., deviation bar, dial or circular bar). Scatter diagrams focus on the relationship between data series (or variables), while box-and-whisker charts and histograms profile the distribution of the data.
Time Continuum. The time element is certainly a key dimension when selecting the appropriate graph style. The choice here is quite simple. The data series is either captured at a point in time or at several points over time. The strength of the pie, bubble, stacked bar and mosaic graphs is their ability to simply visualize all the detailed data from one slice of time. Side-by-sides also allow two alternative timeframes to be displayed and compared. Alternatively, histograms scatter charts and time series graphs excel in analyzing data over multiple time periods.
Number of Data Series: A distinct trade-off typically exists between the time continuum and the number of data series that can be displayed. For a slice in time, multiple data series can be easily displayed via bar, pie, area, radar and bubble graphs. Cluster range, change bar and paired bar graphs succinctly capture the difference between two data series. Multiple time frames with at least three or more data series require specialized graphs such as bubble-over-time, trilinear or time variant cluster bars.
Data Metrics. It is also important to consider what data format is most appropriate for displaying the business trends. Although most original data is linear in nature, derived data such as ratios, indexes, percents, logarithmics and cumulatives can often be more insightful as KPIs. For example, in situations where severe differences exist in line-of-business growth rates or in the actual size of businesses, logarithmic scales temper the scale problem and provide a more consistent read on the business situation.
The graph selection road map in Figure 1 profiles each of the graph styles by the aforementioned attributes and provides a logical starting point for matching business need to the appropriate visual graph presentation. As an example, if we want to compare cost elements across the business at a point in time, the cluster bar graph would be an appropriate selection. Alternatively, costs could be compared with revenue across time by displaying a bubble within bubble chart.
Figure 1: Graph Selection Road Map
Keep in mind that further impact and analysis power can be created by combining multiple graph types into a comprehensive story line. For example, a stacked bar and line graph could be combined to illustrate not only current year sales revenue by line of business, but also total sales compared to a trended monthly rolling average or year-ago totals. Rather than just point-in-time data, this combination of graphs properly frames the current sales data by adding some level of context. Remember, however, that selection of the wrong graph (or the wrong combination of graphs) can be more dangerous than graph omission if the wrong message is conveyed.
Next month, we will show how a Six Sigma technique - Design of Experiments - can assist us in developing the optimum CPM dashboard design to meet the varying business requirements of the user community.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access