Creating an effective Dashboard from raw data is the result of a comprehensive business intelligence process, typically based on three pillars: gathering requirements, defining Key Performance Indicators (KPIs), and building a data model. However, the relevance of the design is of paramount importance. A poorly designed Dashboard may fail to convey useful information and might even make the data less understandable than it was originally.
How to Recognize a Well-Designed Dashboard
-
Simplifying Complexity: We deal with a vast amount of information, ever-changing data, and diverse analytics needs. Our goal is to simplify all this complexity.
-
Telling a Clear Story: We aim to connect data to its context within the business and address user inquiries. This is where the visual presentation of a Dashboard plays a crucial role.
-
Expressing Data Significance: The chosen visualizations should accurately represent the data and the insights you intend to extract.
-
Revealing Details When Necessary: We want each user to access the data they need, neither more nor less. Some users may require a granular view of the data, while others may settle for an overview.
While requirements, constraints, and objectives may differ for each data Dashboard, some general principles apply to almost all cases. We'll present four of these principles and explain how to apply them to your Dashboards.
First, let's examine a poorly designed Dashboard and the design flaws that are immediately noticeable:
-
An excessive number of widgets (around 30), visualizations, and indicators clutter the view.
-
Basic questions like "What is the total sales amount?" take more than five seconds to answer.
-
The visual presentation lacks any organizational principles; widgets seem randomly placed.
-
The lower tables provide little information.
So, let's summarize these principles through four rules.
1. The Five-Second Rule
Your Dashboard should provide relevant information in about five seconds. It should be able to answer your most common business questions at a glance. If it takes you several minutes to find the desired information, the visual presentation of your Dashboard may be problematic. When designing a Dashboard, try to follow the five-second rule: it's the time that should be sufficient for both you and your stakeholders to find the information you're looking for. Ad hoc analysis will naturally take longer, but the most critical measures, those most frequently needed by Dashboard users during their workday, should stand out immediately.
2. Logical Presentation: The Inverted Pyramid
Display the most significant information at the top of the Dashboard, followed by trends in the middle and granular details at the bottom.
When designing a Dashboard, it's important to adopt a form of organization. One of the most useful is the inverted pyramid (see image).
This concept, derived from the field of journalism, organizes the content of a news bulletin in three levels of importance: primary information is at the top, followed by significant details that help understand the overview, and then general or contextual information, which is much more detailed and allows readers to delve deeper. (Think of this structure as similar to a newspaper article, with a headline, subhead, and body text.)
Why can a journalistic technique help design a Dashboard? Because business intelligence Dashboards, like news bulletins, tell a story. This story must follow the same internal logic: keeping the most critical information at the top, trends (which place this information in context) just below, and more granular details (which users can explore further) at the bottom.
This method applies to all contexts (strategic, analytical, or operational).
3. Minimalism: Less is More
Each Dashboard should contain five to six visualizations, no more. Some designers feel the need to saturate their Dashboard with details to offer a richer view. Even though this approach may seem theoretically sound, cognitive psychology teaches us that the human brain can only absorb about seven, plus or minus two, elements at a time. Therefore, that's the number of elements your Dashboard should contain. If you exceed this limit, visual clutter and noise will distract and divert users from the Dashboard's primary goal.
To avoid visual clutter, you have two options:
-
Overlay the data using filters and hierarchies (e.g., instead of having an indicator for sales in North America and another for sales in South America, provide users with an option to switch the same indicator between these two regions using a filter), or
-
Simply divide your Dashboard into two or more distinct Dashboards.
Example: An overly cluttered and unreadable Dashboard
Clear and readable Dashboard
4. Choosing the Right Data Visualization
We discussed data visualization earlier; in a nutshell, data visualizations are meant to be more than just visually appealing—they should serve a specific purpose and convey facts more effectively than the basic tabular format.
Before selecting a visualization, think about the information you're trying to convey:
-
Relationship: The connection between two or more variables.
-
Comparison: A side-by-side comparison of two or more variables.
-
Composition: Breaking down data into distinct elements.
-
Distribution: Ranges and groupings of values within the data.