Data Dashboards: Substance Over Style

In this blog post I want to try and highlight some of the pit falls that catch all developers out at some time or another.

When it comes to developing a dashboard, I have a slightly tautologous saying: “Doing something simple is really easy, but doing something a bit more complicated is hard”. It sounds stupid I know, but there is some merit to it. Let me explain…

I have been involved in many dashboard implementations, many of them involving Tableau, and the same things come back time and time again. We have all been sold the dream that building great visualisations is child’s play but, at least in a business context, it usually isn’t that straight forward.

It should be clear from the beginning what the business benefit of a dashboarding project is. Is the purpose of the project to centralise the data, therefore gaining efficiencies by reducing the time and cost associated with manual data aggregation and consolidation? Is the purpose to gain improved insight into customer data?

The question is: why are we implementing the solution in the first place? Whatever the reason, it should be defined before kick-off, refined during implementation, and measured afterwards. Any project can be considered a failure if its success criteria are not stated up-front. Management, with input from users, must agree on composition of the dashboard and the project sponsors must ensure that success is measured post-implementation or they risk a lack of user acceptance.

When constructing a dashboard we can all get carried away with the exciting visualisations that we have seen on the internet. What we forget is that these dashboards have often been created backwards: the developer has set out to create something interesting, so has found a dataset that will fit nicely into their design. In the real world we work the other way around; we have some data and we want to create a dashboard with meaningful metrics and actionable insight within it. Using data to create beautiful visualisations is simple; you can choose whatever data you want to match your design, but using visualisation to effectively display the data you have is much harder.

Style over substance or substance over style. If you want a dashboard to work for you, you must prioritise the latter.

See, I told you it made sense.

Jon Ede
Head of Operations & Development, Fuel