Dashboard design efforts can easily run into trouble. Many business intelligence dashboards, meant to present BI data to business users in an easy-to-grasp way, instead end up representing a triumph of form over substance and sizzle over steak. But it doesn't have to be that way.
Too often, BI dashboard designs include "gimmicks that can take away from reading and understanding the data," said Charlie Claxton, vice president of creative strategy at Produxs Inc., a Seattle-based developer of custom Web and mobile applications. For example, one of the most common mistakes Claxton encounters is the use of three-dimensional charts in dashboards. "Even though they might look nice, they make data interpretation very difficult," particularly in pie charts, he said.
Claxton also finds a lot of unnecessary clutter in dashboards. To avoid that, he suggests working to apply the clarity and simplicity that using pen and paper encourages to the dashboard development process. "Since BI is extremely strategic, every element that is part of the dashboard needs to have flow and significance," Claxton said. "Every element should have direct meaning and thus should only be present to convey a message."
Clutter is one of three classic mistakes that dashboard designers make, said Steven A. Lowe, founder and CEO of Innovator LLC, a consulting and custom software development services provider in Chattanooga, Tenn. Cramming an excess of data on a single dashboard page doesn't work, Lowe said. "If you have to read it or study it or scan it repeatedly to find the important bits of information, the dashboard has failed its purpose. All useful information should be discernible at a glance."
Turn off the dashboard data spigot
A cluttered dashboard usually leads to Lowe's other big design missteps: including irrelevant information and presenting data in a chaotic way. Too many dashboards squeeze together "everything including the kitchen-sink historical average drip rate" without any logical grouping of data elements or graphical contrasts, he said.
Another common design mistake, in Lowe's eyes, is viewing a business dashboard as simply a more graphical version of a static report populated with historical data. "A dashboard must convey current, meaningful and actionable information at a glance," he said.
Ensuring that it does depends on successfully gathering data analysis requirements from business users before starting the design process, said Bill Brydges, a managing director at consultancy MorganFranklin in McLean, Va. It's also important to fine-tune dashboard designs to support different levels of analysis functionality for different workers, he added.
Many designers mistakenly "assume that all users consume data in the same manner," Brydges said. "Make sure you know your audience. An analytical exec who likes to get behind the numbers will be frustrated by a summary dashboard that doesn't support drill-down." Conversely, a great deal of effort can be wasted building capabilities to drill deeply into data for business managers who only want to see red-yellow-green traffic light indicators on key data elements to alert them to issues requiring their attention.
In addition, many executives and other users are already inundated with data. Designing BI dashboards that include a broad swath of corporate data and are aimed at end users in various roles only adds to the information glut, Brydges said.
Start low, then grow on dashboard designs
"Rather than trying to craft a master dashboard that encompasses cross-enterprise information in a single place -- a top-down approach -- start with lower-level dashboards that focus on a single specific item," he said. Then work to consolidate the information in different ways for individual executives and managers, based on what data is relevant to them.
Lowe agreed that dashboard users should be involved from day one of the design process. But he said designers shouldn't shy away from thoroughly questioning users about the planned uses of the data they're asking for and why it's important that they see particular pieces of information. "It's easy to put a gauge showing average daily sales on a display," Lowe said -- but that might be a bad move if the data isn't really needed for "moment-to-moment decision making."
Keith Metcalfe, a vice president at BI technical services provider WCI Consulting in Plano, Texas, said dashboard design teams should avoid letting the available technology define their plans. Instead, he advised, they first should document the key performance indicators for their businesses and then shape dashboards to fit the KPI needs.
Data security issues also shouldn't be overlooked when building dashboards, Metcalfe said. Some KPIs and other pieces of data are meant to be seen only by certain people in an organization. The right information needs to be targeted to the right roles to ensure that sensitive data is handled properly, he warned. Not doing that could get BI managers and dashboard designers a lot of unwelcome attention from corporate executives.
Alan R. Earls is a Boston-area freelance writer focused on business and technology.
Follow SearchBusinessAnalytics.com on Twitter: @BizAnalytics_TT