Seven steps to a great dashboard
We have designed and built hundreds of visualization solutions for clients across industries and across the globe. While the solutions we have built has consistently delivered the benefits of greater automation and fewer manual errors with the use of the best-in-class data engineering and technology, the real value has been delivered and acknowledged ONLY when we have combined this with a greater focus on design/aesthetics, choice of KPI, the flow and storyboard (of course in collaboration with the clients, but led by us).
From our perspective, the design and build of a great dashboard requires teamwork — project managers, designers, business analysts, data engineers, developers and testers come together with the client stakeholders to ensure a master-piece is built. Trying to work on such a project with fewer people leads to sub-optimal results. Here are seven key steps to designing and building a great dashboard.
- Start with the audience and objective in mind
We need to be able to identify one user, one small set of users or a persona that you are building a dashboard for to start with. This is important because this will help you design the dashboard with the user’s goals in mind, interview the user(s) to get feedback and fine-tune it to be exactly what that user needs. Designing with too many users in mind will just result something that is not great for anyone and just good-enough for everyone.
One more thing to keep in mind is that while this is about a user (especially in startups), it may so happen that a single individual may be playing multiple roles. We should try not to mix the needs of these two roles into one dashboard.
2. Create a story and a flow
For that single user that you have identified, first understand what their success and failure is best measured by (goals/targets/KPIs) and create a story board that will identify and prioritize their top 3–5 key performance indicators. For each KPI depending on whether it is a target (e.g. Sales) to be achieved, a # to minimize (e.g. Cost) or a ratio to be maintained, we can choose the right visual to represent the KPI.
In addition to showing the 3–5 KPIs, one also needs to show how 1–2 of these KPIs are broken down by key dimensions (e.g. Sales and Profits by Business Unit and Product Line) and these too should be in the dashboard. Once this is designed, the rest of the dashboard flows from these 5–7 visuals that will be the starting point. e.g. If your KPI shows that you have not achieved your targets, drilling down to why would be the next logical step and that should be answered in a separate tab/dashboard and interaction setup suitably.
3. Design and Aesthetics are key
A well-designed and aesthetically appealing dashboard is key to making the dashboard something that users will want to use. While Colors and Fonts consistency with the brand are important elements, spacing of the key widgets to ensure that the dashboard is not very crowded is also important. The key is to make sure that the #s that matter stand-out without the user having to put in a lot of effort.
4. Test and Verify the Numbers
A well-designed and aesthetically appealing dashboard is useful only if the numbers are accurate. Before the dashboards are published to a wider audience, the numbers need to be validated and vetted by comparing it both to the source and to other reporting solutions currently being used. For this, before you start on the development of the dashboards, you will need to ensure that the test-cases for the dashboards are pre-defined, cover all cases including edge-cases and well-documented so that there is clarity on when testing is complete and also so that there is rigor and discipline in testing. When there is a data pipeline in place, building test-cases and test-scripts for each stage are also critical in ensuring that the pipeline is stable so that one fix does not lead to another failure.
5. Choose the right drill-downs
Each choice of KPI should be followed by an answer of why a KPI is met / not met. Our rule of thumb is that if not more than 3-5 clicks should be needed to get from question to insight. If this happens, the solution is probably to break this into a different dashboard.
For example, I see that my company has not hit the Revenue # for this quarter. The time-period is Quarter to Date, the benchmark is Target Set and Possibly a QoQ/YoY #. Now, the possibilities include the under-performance of a Sales Rep, a Geography (all reps within the Geo), an existing account or just an aggressive target that may not have been realistic. The visuals should be grouped and the flow created such that the answer to the why should be available in max 2–3 clicks.
6. Choose the right KPIs and the right visuals for the right KPIs
Since you already know who the audience / persona you are building the dashboard for, spend time with the audience to identify and prioritize the KPIs that matter and understand how that KPI is achieved / missed and what the benchmarks / comparisons for it are.
Depending on the KPI being a #, a ratio/percentage or being a $ figure, the choice of the right visual (including the style, size, color, conditional formatting, background, spacing) along with the right use of spacing and options the widget comes with will ensure that one look and the insight pops out.
7. Choose the right time-periods for the right KPIs
Each dashboard is ideally defined for a point in time or for a specific time-period. Trying to confuse users by giving them too many options and too much flexibility within one dashboard can cause a poor UX leading to overall poor usage. A point in time dashboard (e.g. Stock Report) should not be combined with a Report that looks at a KPI within a given time-period (Production Report). That is where clear delineation of these into separate sections or better still a separate dashboard will result in a better UX. Similarly, combining a QTD viz with a WTD or MTD viz also typically needs to be avoided as it leads to unnecessary context switching.
Do follow these in your next project and share your feedback on if any of these help. Of course, if you need help on your next project, do reach out to us at www.convergytics.net or connect with me on LinkedIn.
Author Profile:
Randhir Hebbar is an entrepreneur and one of the founders of Convergytics — Asia’s leading analytics brand (as per UK based Global Brands Magazine). Randhir heads Data, Digital, BI and Products at Convergytics and manages several key accounts. He has consulted with dozens of leading Fortune 500 brands over the past 20 years.
