“Numbers have an important story to tell. They rely on you to give them a clear and convincing voice.” ― Stephen Few
The dashboard design is most of the times considered as the cherry on top of the cake on a BI Project, in fact, it is one of the main factors that define the success of a BI project since it has the biggest impact on the end-user experience.
You have completed the biggest part of your work, you have collected data from different data sources, cleaned it, applied business rules, created useful metrics and finally placed this data into a beautifully designed Data Warehouse. Now it is time to turn data into information, and information into insights. However, how do we define a clear, useful and user-friendly dashboard? Below, I am sharing some guidelines I have learned from my experience implementing BI solutions for large companies:
1. Know your audience
Defining the audience is the first and most important step in designing an effective dashboard. It is very important to make sure that the dashboard’s audience will relate with it. For that, you need to clearly define who will be looking to the dashboard, what they are looking for, and in what context they will be using it. Without having this clear, you cannot deliver an effective dashboard.
Is the dashboard going to be used by the senior leadership to monitor the company top line figures or will it be used by the sales representative to monitor daily activities? It is important to ensure that your dashboard consists of data that is specific to a single audience.
2. Classify your dashboard
Understanding which type of dashboard you are designing will guide you on your design decisions. There are three common types of dashboards, each one with its purpose:
> Operational: Displays data that facilitates the daily operations of the business by managing core KPI’s (Key Performance Indicators) of the business process. The objective is to help a department understand if its performance is on or off target and by how much. Most of the times, operational dashboards use real or near real time.
> Strategic or Executive: Delivers the necessary KPI’s to allow the companies’ leadership to track the performance on a periodic basis. Normally it provides a high-level overview on the current state of the business along with its objectives.
> Analytical: Even though it can show strategic or operational data, it provides the user the capability to explore the data and get different insights. Examples can be comparing trends over time or combining Sales and Marketing data to determine the success of individual campaigns.
3. Choose the right visual
Data visualizations are essential for an effective dashboard, since people find it easy to process information in a visual format. However, making the wrong choice on the data visualization format may cause misunderstandings to the end-user, potentially causing bad business decisions. The list below is not exhaustive but intends to summarize the most important chart types and its purpose:
> Indicator: To display a single value. It is important to label the period correctly and give a reference number;
> Line Charts: Ideal when checking (one or more) measures over a period of time. By understanding how data changes along the time, we can identify trends and patterns that can be interesting;
> Bar Charts: Good when the objective is to compare values across same categories (country, product, customer);
> Combo Charts: Combines the features of line and bar chart, representing different measures in the same visualization;
> Pie Charts: Useful when you want to see the contribution of the parts in the whole. Since the pie chart represents the size relationship between the parts and the entire entity, the parts need to sum to a meaningful whole. Although it is one of the most popular, as it is very easy to understand, it is not very consensual within the BI world, because of the lack of accuracy, especially on the smaller values of the pie. The common recommendation is to use pie chart with three or less segments;
> Stacked Bar Charts: Multiple part-of-a-whole representation, interesting to compare many different items and show the composition of each item being compared;
> Scatter Plots: Show the distribution and relationship of two data items / measures;
> Bubble Charts: Represent the distribution and relationship of three data items / measures.
> Tables: Ideal for values lookup and detailed analysis.
4. Provide context
Every element of a dashboard should have a context for the numbers being shown. Without comparison values, values are meaningful for the audience. Otherwise how can the end-user understand whether a value is normal or unusual? Try to use comparisons that are most common, for example, current period against prior period, or against a target or expected value.
5. Take care of the details
One of the best practices most mentioned is to try to combine the most important information into a single screen, you should try to avoid over flooding your audience with information. If that is the case, try to create sheets/tabs for different themes/subjects. You can split a finance dashboard into sections like you would do on a website.
Some other recommendations:
> Never forget why you are doing the dashboard, and try to do it as simple as possible;
> Put yourself on the user’s position;
> Consider how the dashboard will be viewed (context and device). Try to make it as much responsive as possible;
> Start with the big picture, key information should be at the top left corner.
6. Use visual features properly
> Always place time on the horizontal axis;
> For numerical bar charts, numeric axis should start at zero;
> Label necessary items (titles, axis, units, series). Do it smartly, especially if you have space constraints (ex: mobile devices);
> Avoid 3D charts, and unnecessary patterns (ex: backgrounds, borders) which may obscure the actual data;
> Use white/blank space between the elements, to make them readable;
> Choose a few colors and stick to them, if necessary use hues of the same color;
> Use simple fonts (Arial, Tahoma, etc.).