BUILDING EFFECTIVE DASHBOARDS
Dashboards are a much-loved tool in analytics and wider business. If they are done effectively, they can keep the business up to date with key figures, prompt areas for further analysis, drive decisions and entertain a curious mind. And of course, look very pretty.
A large percentage of dashboard developments fail in terms of any sustainable success, however. They either end up lying dormant, unseen and misunderstood, or eroding into chaos like an unkempt garden. Ghost dashboards you could say.
Such failures are unnecessary. Here a few of my tips to ensure dashboards can successfully be used to drive the key decisions that can transform organisations.
1. Fully understand end users requirements
The end user is supreme. This is the golden rule above all. A dashboard may be brilliant and beautiful in all ways but if it does not help users do their jobs it is merely unnecessary CO2 emissions. Developers must fully understand these requirements, what needs to be reported and what decisions needs to come from it. Developers must probe, probe and probe again. Often the users need help, and pragmatic analysts and developers will help them understand their needs, open up the world of possibilities and limiting unrealistic expectations.
2. Fully understand end users access and abilities
It is one thing meeting requirements but a dashboard must also be understood and accessible. Developers should recognise users’ technical abilities, their understanding of business terminology, where and when they work. This will allow a dashboard to be built that is easy to use, set out logically and updated at appropriate times. A big win for a dashboard is to be no more than a click or two away and knowing the users’ set up will facilitate this. And of course, fulling training with cases studies should be rolled out with any implementation, this will not only educate the user but also expose any missing gaps.
3. Understand the data
This goes without saying of course but understanding the strengths and limitations of data is critical to success. Data will be coming with different sources, at different times and in different formats and this must be recognised. Work in pre-development will allow developers to know exactly what data is needed, the onus is on the developer not to be tempted to overshoot on this.
4. Compartmentalise the data and the dashboard
Separating data manipulation from the dashboard itself is key for a sustainable development. An obvious benefit will be increased speed with the dashboard. Specialist coding tools are also much better at manipulating data. But more vital is changing data or business requirements. If these change the data processing or dashboard can be amended independently which thus allow much great adaptability and flexibility.
5. Don’t over-automate
Too often I see a focus on full automation at the expense of other factors. This can at times be achieved effectively but more often it comes at high costs. These costs include having to exclude some key data sources, toleration of unclean data, production errors, limited usability and other factors which inhibit business use. And of course, full automation usual means inflated development time. Almost always 10-15 minutes work per day/week to intervene, monitor and adapt a process is the better than an automated but flawed and often unusable dashboard.
6. Substance over style
It is important a dashboard is tidy and visually attractive but it does not need to be the Sistine Chapel. Dashboards should allow plenty of interaction and drilling down, but they are not tools for bespoke deep dives. Nor should they be real time when a daily or weekly update does the job. You get the jist - stick to user requirements.
7. Keep tracking and improving
Dashboard tools are fortunately (and creepily) fitted with devices which tell you who is visiting which page and when. Watch this. If certain users are not participating regularly or pages left unvisited find out why. And adapt. The whole point of dashboards is regular usage. And go that step further, have regular discussions with users to get a feel for how they are using the dashboard, it is not just about access but seeing the insights are being used to drive to actual decisions.
This is clearly not a technical discussion. With a bit of experience dashboards themselves are relatively easy to build, the likes of Power BI and Tableau can be mastered well with some training and experience. It is the work at either end that defines success or failure - understanding the users and understanding the data. With that dashboards can indeed be the beautiful tools that transform our organisations for the better.