Have we all become dashboard junkies?

 

As COVID-19 surged, the world has relied on dashboards to tell the story of the pandemic. Refreshing our screens with the click of a mouse, we saw the pandemic unfold on a world map and in dynamic demonstrations of data morphing into charts and graphs. Containing steady streams of timely information gauging testing rates, case counts and the ensuing death toll, dashboards have been used to both inform the general public and to guide decision-makers to manage the pandemic.

By synthesizing large quantities of data into an accessible snapshot of the situation, distilling insights and making this knowledge readily available, dashboards are a beguiling communication tool. The pandemic has exalted the idea of a dashboard, which has evolved from a speedometer plus a few warning lights, to something that would require concentration rivalling that required to fly a stealth fighter-jet. The technology is fascinating and seductive. “We can do it” slicing and dicing data in all sorts of ways we never could have before. We can make data bounce, flash, zoom and soar. We can uncover covert relationships at the push of a button, play out scenarios, peel layer after layer to attempt to understand what is really going on.

But not all dashboards are created equally. Like any communication tool, dashboards must be designed for their audience, who have different types and levels of engagement. An overview page for example, should be widely appealing, leading the way into more detailed data for those who are interested. A busy manager may need explanatory analysis – information to help make decisions, guide work and to inform advocacy. As one of my colleagues put it recently, “Managers say they want data but they actually want information”. They don’t want to be spending time delving deeply into an expanse of data or navigating between the roundabouts of pie-charts and the helter-skelters of bar charts in order to find out what’s important. This calls for a dashboard that’s easy to navigate and a simplified design with readily available insights. A researcher on the other hand, may take interest in the process of exploring and require a more granular view of the data with options to drill down.

Dashboards and the evidence they provide can also be misinterpreted or misused, spreading through digital channels like wildfire and creating a crisis of misinformation. The recent use of data that were outdated and overestimated deaths, to justify England’s second lockdown, drew criticism from regulators. It is essential that we hold the public trust with care – we must wield these tools with caution and be transparent about how we derive the data we use. Though the best dashboards present data in a way that’s as truthful and as objective as possible, no dashboard is entirely independent of its maker. At offer isn’t a dataset alone, but a perspective that colours the information presented. Every viewpoint, every choice of filter, and the hierarchy established have underlying ideas or motives. Meaning is even codified within a choice of colours which can signal different messages from alarmism to neutrality to prosperity.

Ideally, a dashboard will offer an analysis that’s cross-cutting, showing relationships across buckets of information that aren’t readily visible in a single chart or graph alone, leading the user from data to information and from information to insights. This is particularly essential for dashboards that are intended to inform decision-making.

Either way, for many during the pandemic, dashboards are the primary source of gauging the virus’s reach and impact.  As we are ultimately reliant on these tools, let’s use them intentionally, and let’s bear in mind a few simple guidelines to place our users’ needs first. I offer the modest list below, compiled by another colleague:

 

Dos

  1. Keep it relevant – your dashboard should be accessed easily by your audience. This will call for different designs for different audiences. For example, for busy decision-makers, the overview page should contain key insights that are simply and clearly laid out.
  2. Use charts to summarize complex information into easily digestible information and try to group them together or place them logically.
  3. Pool data from multiple channels to present a full picture.
  4. Use icons or legends to guide your viewer through a dataset/graph.
  5. Place relevant descriptions and/or captions wherever can be helpful for the viewer.
  6. Add filters to make the dashboard more interactive.
  7. Beware of distractions to avoid confusion such as too many graphs and diagrams or too many colours.
  8. Take into account robust testing and future maintenance.

 

Don’ts

  1. Don’t assume that your audience knows where to start when viewing.
  2. Don’t overdesign – there’s no need to place all the information on the same page.
  3. Don’t use too many colours – don’t use gradients, analogue gauges and 3D visuals in graphs.
  4. Don’t add many indicators on a single chart, creating a complex dashboard.
  5. Don’t try to answer every question at once.