Input: What is a Dashboard?

I have great admiration for research projects that set out to investigate seemingly obvious things, and then suddenly a whole ocean of questions opens up. “What Do We Talk About When We Talk About Dashboards?” 

This brilliant question was posed by a group of five authors in a paper presented at IEEE vis, the big visualisation research conference which took place in Berlin last fall. So yes, what is that – a “dashboard”? Here’s my executive summary of the paper. 


DB005 from the study’s example corpus: Shine Pulikathara, 50 Years of Crime in the US

What’s this all about? 

We all have a more or less vague understanding of what a dashboard is. This vague understanding probably includes a visual scheme composed of several graphs and diagrams and a few key metrics, which is either static or refreshes automatically from time to time. Sometimes, a few interactive features will be involved. So far so evident.

Now, over the last years, dashboards have made a career to become one of the major interface types through which users interact with “big data” (read: large structured digital data sets). Likewise, use cases and contexts of use for dashboards have expanded significantly over the past years. Despite this, the prolific visualization research community (most of which is rooted in the computer sciences) to date has largely ignored the existence of the dashboard as a type of complex visualisation that is ubiquitous in many industries.

This observation formed the starting point for a research project by the self-proclaimed “Dashboard conspiracy”, formed by five authors from Tableau Research, Microsoft Research and Simon Fraser University near Vancouver, Canada. They wanted to understand exactly what a dashboard is, which major forms can be observed, and which open questions for future research can be derived from current use cases.

How did they go about it?

The authors used a double approach. First, they collected a total of 83 dashboard samples out in the wild. They used this sample set to identify several major types of dashboards. These were their criteria for describing the visual and functional features:

    1. PURPOSE (The roles of each dashboard in the process of analysis and communication)
      Decision support (Yes/No)
      Communication and learning (Yes/No)
      Circulation (Public, Social, Organisational, Individual)
      Required visualisation literacy (High, Medium, Low)
      Domain expert knowledge required (Yes/No)
      Construction & Composition (users can design and/or customize a dashboard)
      Multipage (users can create tabbed views)
      Interactive Interface (users can create multiple coordinated views, using slicers, filters, drop-down, cross-highlighting etc.)
      Highlighting & Annotating (users can annotate or highlight specific views)
      Modify Data or World (users can write back to the data source OR influence factors outside of data set)
      Alerting & Notification (dashboard identifies anomalies and critical scenarios)
      Benchmarks (dashboard shows goals, directions, thresholds or status lights)
      Updatable dashboard (refreshes automatically, as opposed to drawing from a fix dataset)

Secondly, the paper reviews a vast array of literature tackling the creation and use of dashboards, coming from a variety of professional fields, from business intelligence to health care management to non-profit organisations to urban informatics.

DB014 from sample corpus: Ken Flerlage, All My Vizzes

So what did the authors find out?

One pivotal result of both the survey and the literature review was the observation that the term dashboard references both a visual design idea of what that should look like (i.e. tiled layouts of simple charts) as well as a functional idea of what a dashboard facilitates (i.e. to support decision-making or to educate/communicate). These two are not always consistent—sometimes a complex visualisation will function like a dashboard in that it supports monitoring and data-driven decision-making, but not look like a dashboard etc. Therefore, the authors conclude

“that the term ‘dashboard’ does not pick out a unique method of organizing, presenting, and using data, but rather covers a diverse set of practices.” [§6.1]

Furthermore, the authors state that the use of dashboard not only proliferates, but that their use cases and contexts are extending into “nearly every industry, non-profit and service organisation to support data-driven decision making.” From the collected sample of 83 dashboards, they identified seven types of dashboards:

    1. Dashboards for strategic decision-making (mostly for organisations)
    2. Quantified-self dashboards for motivation (mostly for individuals)
    3. Static dashboards for operational awareness (data from sensors, low visualization literacy but high domain knowledge, no interactivity, mostly for organisations)
    4. Static dashboards for awareness (for reporting static at-a-glance overviews, mostly for organisations)
    5. Dashboards for operational decision-making (e.g. up-to-date business metrics about sales, mostly for organisations)
    6. Dashboards for communication (to educate about public health, crime rates etc., mostly for the public)
    7. Everything else that didn’t fit before (many looked like dashboards, but didn’t function this way)
Overview of the seven types of dashboards

So what follows from this? 

First of all, I thought it was extremely helpful to look into dashboards as a visual and functional scheme for interacting with complex data sets. What are their features? What are they used for? More specifically, I learned from the study how dashboards have evolved from a mere monitoring device to a more complex tool for fostering decision-making based on data:

“As the scope of dashboard use moves from merely reporting performance via proxy metrics to more in-depth problem solving, users also want more analytical support in their tools, particularly ‘smart’ features such as automated analytics (pattern identification, prediction) and what-if simulations, or engagement features such as goal-setting and gamification. Automatic classification is one feature already showing benefits, such as automatically classifying a user’s location (e.g., home, work) from GPS data in personal monitoring dashboards.” [§5.2.1.]

Among the latest developments is the fact that increasingly the format “dashboard” is hacked for communicating data sets to non-experts – which requires different approaches to their design (e.g., more annotation). From these observations, the “dashboard-conspiracy” authors derive a number of current challenges in the field of dashboard design, such as:

    • more reflection about the data choices (the choice of specific metrics mirrors priorities)
    • the need for more sophisticated metrics for solving complex problems (impoverished data)
    • more options to adapt dashboards for various user tasks
    • an increasing need to include metadata into dashboards (specifically for revealing data quality)

Also, the authors hope for a more mature handling and dealing with the data displayed, both from the users side as well as the designers side. Issues of privacy, trust in the data, notions of objectivity and truth etc. are discussed in the last section of the study.

The paper “What Do We Talk About When We Talk About Dashboards?” was presented at IEEE vis in Berlin on Oct. 25, 2018 and is available here. It is also worth listening to the Data Stories episode “The Dashboard Conspiracy” dedicated to the topic with Alper Sarikaya and Lyn Bartram, two of the five authors of the paper.  

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s