I first became wary of automated data dashboards when I was told I could receive my reports in a daily email. The more emails I receive from the same sender, the less likely I am to read them, or even notice they are there.
When did you last see a no smoking sign? It was probably today, but they are so common you don’t remember. Similarly, greater frequency of reporting does not necessarily equal greater value.
The dashboard is nothing new, and summary displays have long been common to present large amounts of information. A key difference today is the fairly recent rise of very powerful visualisation tools, such as PowerBI, Tableau, and RShiny. These present information in an aesthetically pleasing and apparently convenient display. But users can get carried away, with over-use and over-reliance resulting in mechanical thinking and fully-automated reporting.
This should be a red flag.
The secrets to avoiding the pitfalls of automation are good design and proper usage. By being clear about what the dashboard can and can’t do, and understanding the picture you are painting, you can create better tools that add value and improve processes.
Choose quality over quantity
Just because you can have daily updates, doesn’t mean you should. Before implementing any sort of automated cover you should be thinking about what the user is going to do with the information.
Start with the extremes. If you saw that everything was as high or low as it could be, what would you do? If the answer is ‘nothing’, stop now. If you would act, ask how much less extreme would it have to be for it to be okay?
Following that, consider whether the insight is likely to result in action, or numb your sensitivity to triggers. This is where understanding the quality of your output is vital.
Take care when changing parts
If you want to change your dashboard, do you have to worry about changing inputs and results? Do you have to change it for each output? The answer here should always be no. One change should flow throughout, and you shouldn’t need to hit copy and paste. If you’re setting things in stone so that this isn’t possible you need to reassess your game plan, whether it be for a small project or a large one – as every large project starts small. If it’s not scalable, it’s not suitable.
In life, beauty is in our differences. The same cannot be said for data visualisations! When things look similar, people assume they are the same. Be wary of making minor tweaks between different versions in your dashboard. If you start tweaking different, but similar, dashboards for different uses you’re sure to confuse your reader sooner rather than later.
Let the human take the wheel
Dashboards that meet the criteria above can be very useful tools. They allow a quick and timely condition assessment to be made. But, fundamentally, reports are a valuable opportunity to tell your story, and as such they are a place where a human should step in and highlight the ‘Yes, buts’ and ‘Howevers’. Without this human awareness, mechanical thinking can lead us to do things the way we have always done them, for no other reason than ‘that’s the way they’ve always been done’.
If a dashboard offers a creative approach to quickly giving everyone the information they wanted to know, like a driver’s speed as they change speed limit, it’s a good thing. But if it’s replacing an opportunity to assess how well they’re really driving, someone needs to put their foot on the break – sharpish.