Six Questions You Need to Ask Before Building a Dashboard
Modern business intelligence software has laid the static PDF or Excel reports to rest, and now lets users interact with their data through dynamic BI dashboards. However, building the perfect dashboard can be tricky, especially with the multitude of visualization and design possibilities that are usually available. A dashboard that is not designed correctly will fail to convey the relevant information or fail to convey it clearly, and hence will not produce any real value in terms of aiding decision-makers and providing a bird’s-eye view of the business (or the department, or a certain process that is being examined, etc.)
There are many factors that need to be taken into account when deciding which elements to include in a dashboard. Some considerations are more relevant for certain industries or processes, and hence it is quite difficult to give specific guidelines that will be suitable for every contingency. However, before setting out to create any kind of dashboard there are certain questions you need to ask to understand the ‘story’ your dashboard is going to tell, how that story can be told as clearly as possible, and who is the target audience.
So without further ado, here are the questions you need to ask to determine how to build your next dashboard. You can also download this guide here (PDF).
What type of decision needs to be made?
In other words: what is this dashboard meant to do? Is it supposed to help executives understand an established process and prescribe results or to explore a new course of action?
For prescriptive decisions, you’ll want use clear KPIs and easy-to-understand visualizations. You should ensure conclusions can be drawn quickly from the visual interface itself and by applying filters.
To support exploratory decisions you need to give the end-user more freedom, so provide a broad set of data and filters. Allow the user to segment the data to create different views of the same data set. You will also want to enable drilling down into details and across segments of data.
What is the primary focus of your dashboard?
Is the dashboard for a specific task, such as overviewing the results of a new project, or is meant to achieve a broader goal, e.g. to measure organizational performance?
If the focus is specific, you want to keep your dashboard simple: use only a few visualizations and standard KPIs or metrics. You might also want to use color alerts to denote whether specific goals are met.
If the focus is broad, your dashboard will also need to be somewhat ‘broader’, accordingly. This means you’ll be using more widgets, filters, and drills. It’s a good idea to group similar sets of charts together and show relationships between different parts of the data.
Is your dashboard retrospective or real-time?
Dashboards could be retrospective — examine a process or initiative that has already concluded, based on historical data; or they could be looking at present, real-time outcomes.
For historical periods, you need to determine how far back the data should run to provide a useful view of the process being examined. This type of dashboard should reflect long-term trends and enable comparisons between time periods.
For the present period, you’ll want to include thresholds and highlight outliers. This type of dashboards is often used to monitor decision-making.
How detailed does your dashboard need to be?
If you’re only looking to reflect trends and general fluctuations within your data, you don’t necessarily need to enable the end-user to drill into the more granular levels of the data, and a static ‘snapshot’ of it will do. But if you want to present a complete picture you will want to allow this type of drilling into the fine details of the data.
For static detail, your goal should be to pre-summarize the data to the degree needed by the end-users. This type of dashboard is meant for users who do not have the need or willingness to delve too deep, which is why you should use standards KPIs and visualizations.
For granular detail, you need to ensure that all the data is available so users can drill into each segment and across different segments of the data as needed. For obvious reasons this dashboard will usually include much larger amounts of data.
Do you need to segment the data?
Again, the question here is to what degree the end-user will actually want or need to go into the ‘bowels’ of the data. Does he or she need to be able to filter and customize the view of the data, or is a fixed view sufficient?
In cases of fixed segments, you want to ensure common KPIs and filters are available. Make the dashboard as clear as possible — only use the data or visualizations that are absolutely necessary to tell the story.
For a customized segment, provide a broad set of data and filters. Ensure users can drill into and across broad segments at will.
How savvy is the end-user?
Will the end-user be familiar with the data and business domain or are they new to the data?
For familiar users, data can be presented ‘as is’ — just ensure that the metrics and visualizations align with the standard way the business is using the data.
For new users, make sure to give the visualizations and fields intuitive titles, and to align the metric to familiar business goals. This dashboard should include more textual explanations of the data it presents.
Want to learn more about dashboard design?
Watch this free on-demand webinar on How to Build a Better Dashboard to discover how you can:
- Understand the big picture for analysis
- Identify critical factors for planning
- Learn best design practice
- Always choose the best chart to visualize data