Telling Stories With Data

I have been thinking about dashboards a lot recently – maybe I have been perusing the Euros 2020 app a bit too much (the data they hold about matches and players is fascinating).

The key to a good dashboard is how fast somebody looking at it can pick up the underlying message – it isn’t just about displaying data in pretty charts.

Thinking about the dashboard I look at every day, in my car, I want to see whether I need to stop to get fuel today – I am not interested in how much fuel I had this time last year and what my average fuel level was over the last 30 days. I don’t have much time to look, because I am driving, so a little picture of a gauge showing me how full my tank is, and if things get really bad, a red light flashing is perfect.

There is no doubt what the message is behind this award-winning dashboard created by Simon Scarr and appeared in the South China Morning Post in 2011:

It shows the deaths by month during the Iraq war. It’s powerful, isn’t it? The blood-like red bars dripping down the page really emphasise what we are looking at. We immediately know that conflict resulted in a huge loss of life.

In designing this chart Simon made a few decisions to highlight the message:

  1. The bar chart is upside-down compared to how we usually see them
  2. The blood-like colour red he used
  3. The title of the chart is evocative

Now, let’s look at the same data presented in the default way a tool like Power BI uses:

Now we have the bar chart the usual way up, and a more neutral colour and suddenly the message changes to a more positive one – we can see light at the end of the tunnel because the number of deaths is declining.

This is the same data – three things have changed:

  1. The y axis is in the more usual orientation with zero at the bottom
  2. The colour is a neutral one
  3. The title is less evocative

I started these ramblings by saying a good dashboard shouldn’t need a lot of scrutiny for the viewer to pick up the underlying message.

Next time you jump to a conclusion based on a dashboard you have seen ask yourself: did the developer have an agenda? What would this graph look like if I change the colour, or reverse the order of the axis values?

And, if you don’t jump to a conclusion the next time you look at a dashboard – get a new dashboard!