Design Principles for Dashboards

Discussion

After reviewing the literature on dashboard design principles and applying these recommended principles to create performance dashboards for quality of diabetes care for the Health Bureau of the Taichung City government, the focus group modified these principles and proposed the memorable 5S principles: 1) seeing both the forest and trees, 2) simplicity through self-selection, 3) simplicity through significance, 4) simplicity through synthesis, and 5) storytelling. The 5S principles provide a helpful guide for non-IT professionals and nonexpert casual users to design effective dashboards by adopting SSBI tools.

One of the contributions of this study was the systematic review of the literature on dashboard design principles with a summary of crucial concepts in Section 3. Because different authors have used different terms for similar principles and techniques, the focus group proposed an easily memorized mantra for dashboard designers to follow. They thus mimicked the 5S method of workplace organization to proposed the 5S dashboard design principles, which can help non-IT professionals and nonexpert casual users to efficiently and effectively design dashboards that can provide insightful and actionable information for decision makers.

Two other crucial contributions of this study were the third and fourth principles (simplicity through significance and simplicity through synthesis), which have not been mentioned in the literature. Unlike those in previous studies, most participants of our focus group were not IT experts; thus, they could provide different points of view without referencing the domain knowledge and daily decision-making required by IT professionals. The two newly proposed principles are intended to help solve decision-making problems encountered by health care professionals in the public sector. We believe that these decision-making problems (conflicting results from excessive and discordant indicators) are also common in other domains.

Another contribution was the proposal of storytelling as a principle. In the dashboards designed by the two participants mentioned in Section 5.5, more specific, prosaic options were composed in the navigation bar across the top of the display such as "identify districts with the lowest checkup rate," "identify clinics with the lowest checkup rate," "location of eye clinics that can transfer for eye examination," and "why did physicians did not check urinary microalbumin?" to help users more clearly understand the plots of their desired stories. However, Figures 2–4 reveal only generic descriptions such as "city comparison," district comparison," and "traffic light ranking". The use of text and annotation can help users to fully understand the context of the story.

Watson noted that many newer visualizations using sophisticated coding techniques are difficult to interpret, especially for senior executers. He reminded designers to ask themselves "are we creating visualizations that are too difficult to understand?" He further suggested introducing a new job, "data interpreter," who explains visualizations to executives, line-of-business managers, and other users who require help interpreting them. The data interpreter must understand the audience, business domain, problem that the visualization is meant to address, and visualization itself, in addition to having storytelling ability and excellent communication skills.

Simply knowing the 5S principles cannot guarantee the creation of effective dashboards. Many people have read the classic style manual The Elements of Style by William Strunk, Jr. and E. B. White, but not all of them can write good articles. Repeated practice and feedback are essential for learning a skill. Another crucial factor in designing effective dashboards is, as Berinato suggested, observing other people's work for ideas and borrowing inspiration. He recommended visiting websites with strong visualization, such as The New York Times or The Economist; the focus group added the Tableau Public Gallery to this list.

Several limitations are worth noting. First, the literature review might not have been exhaustive. However, we already found theoretical saturation of most principles (i.e., many principles repeated in different articles). Second, principles always require a balance between generalizability and specificity. The focus group strove to find this balance. Third, the data we used in this study were relatively simple. However, we believe that many problems encountered by health care workers in their daily decision-making are similar in different professional fields. The 5S principles could still be applied to other domains.

In conclusion, the 5S dashboard design principles are easily memorized and practical principles and enable non-IT professionals and nonexpert casual users to design insightful dashboards efficiently by using SSBI tools.