top of page
  • Paul Verweij

Why your dashboard doesn't work (and how you can make it work!)

Man worried dashboard

The rhetorical force of Covid-19 graphs

If the year 2020 has brought us something good, it must have been the massive exposure of data and visualizations to a worldwide public. All newspapers, news blogs, Twitter, Facebook, were exploding with graphs and tables on Covid-19 data. Many ways to explore the data were communicated to a wide variety of audiences. A complete range of exotic tools and methods were used to create a sheer volume of visual designs trying to convince the common public.

On the other hand, this plethora of charts and graphs that were produced made it very clear that abundance of both data and visualization techniques do not guarantee immediate success or a lasting relationship.

In the end (is it really the end?) only a few graphs seem to have survived the virus. Some put together in a dashboard, some as story telling or scrollytelling pages in news bulletins. But most graphs have just disappeared. Apparently they did not serve the purpose.

The T-Rex of the Covid-19 graphs must be the ‘flattening the curve’ graph. Everybody has seen this one, and it was definitely playing a significant role in driving public action on an international scale. Less than 12 months later even this chart is obsolete, it is in the people’s heads, it worked.

Flattening the curve graph
Source: entomologytoday

Maybe even more intuitive and meaningful is the geographical chart, showing the spread of the Covid-19 virus. This choropleth map is still used in the news every day. Simple and effective. It won the battle with the colourful bubble maps.

Choropleth map Corona spread Belgium
Source: The Brussels Times

Both were a success because they are simple and unambiguous, and they very clearly answer two basic questions: Why do we need to flatten the curve? And how bad is it in my town or country?

Why doesn’t my dashboard answer simple questions?

There are many definitions about what a dashboard should contain, or what it should do, or how it should look. There is however only one thing that really matters: it must start with a meaningful purpose!

The goal of the dashboard, and within it the goals of each individual graph, must lead the way. Without it, the readers will be looking at randomly picked graphs, trivial data and meaningless trends and comparisons.

A complicated factor that notably came to light with the recent Covid-19 dashboards is that the audience was often not known. Because literally the whole world can access and link to the graphs endlessly it is a mere guess whether that dashboard will tell anything useful to the recipient. A dashboard made by (and for) data science students ended up in the newspapers. Another dashboard for medical staff was appointed as a standard for politicians in the Netherlands: unfortunately, it did not work.

Simultaneously a trend emerged where all kinds of people with access to the raw data produced their own graphs and dashboards, habitually without any knowledge of the data and without a sound experience in dashboarding and visualization techniques.

In corporate environments, with their own data and dashboards, these factors can play a comparable role. Before starting to build a dashboard you have to know your audience, know the capabilities of them, their background, their questions and their goals. And of course, you need people that are skilled in dashboard design.

Dashboard design should focus on solving people’s problems. Which visual display best suits the purpose for which they wanted to create them in the first place.

Showing the purpose

It is obvious that a set of data must be visualized to be meaningful. Data, and not only the big volumes, may not tell any story without a visual display. A nice example from some decades ago is the famous Anscombe’s Quartet. It shows that four small sets of numbers, that do not look very special, and that have the same cumulative values like average, sum, deviance; show a completely different view for each set when displayed graphically.

Quartet's summary stats x/y coordinate plane

Quartet's Summary Stats

It is all about visually encoding the data to give it a meaning. In the graphs above you can see that we do not visualize to make it more fancy or appealing to the eye but to be able to turn the data into useful information.

A prevalent caveat is that one may go on a quest to find the ‘best way’ to visualize the data. This will be a disappointing journey to El Dorado: it does not exist. It all depends on the purpose and the audience. One set of data can have different use cases, and for each case another graph will be the best fit. For example in a growing company the recruiters will look at an absolute number of new hires and managers will look at percentual growth. Same data, different purpose.

An additional benefit is that the effectiveness can be measured against the purpose. One cannot measure and quantify if a certain graph or colour is the best way to present the data, but we can measure if: the dashboard answered the questions of the readers, if the audience got the insight they were looking for, if an action or impact could be determined from it.

How can I do it?

Notwithstanding, what common software vendors tell us, it is not the slick graphic design that get people to pay attention. Or only for a short amount of time. The most engaging graphs are simple and not very artistic. You can check for yourself which graphs from the Covid-19 pandemic you will understand and remember.

This one – created by the NVIC and the RIVM to inform a large audience with basic information:

Coronapatients on intensive care in the Netherlands graph
Coronatests the Netherlands graph

Or this one – created by a hobbyist without any guideline in mind:

Corona cases per municipality in Belgium dashboard
Source: covid19-esribelux.hub.arcgis

The rule of thumb is always to keep it simple, with the audience in mind, and designed for insight.

There are simple moves that can make all dashboards or graphs better and ways to make the purpose obvious.

  • Use colours cleverly

  • Limit the number of measures

  • Remove clutter

  • Leave out obsolete and trivial data

  • Highlight individual values that need attention

  • Use charts that are easy to read

  • Compare your data (e.g. against target or average)

  • Use relative data

Bottom line: If you want a useful dashboard or graph, start with the purpose and the audience.

I would like to invite everyone – whose curiosity is now piqued - to contact Jan from CANGURU so that we can go over your specific case. You can send an email to or give him a call on +32 498 91 17 20.

113 views0 comments

Recent Posts

See All


bottom of page