This week’s guest post is by Fernando Bustillo. He is a Business Intelligence and Data Warehouse expert and has experience working with SAP Business Objects, Oracle, Teradata and other enterprise solutions. Contact us if you would like to be one of our guest bloggers.
The fact that many dashboards seem to be a single visual screen or eye candy screen could make one believe that it is easy to manage. This could be the first reason why your project will be unsuccessful: underestimating “the enemy”. In my broad experience working with BI (Business Intelligence) projects, most of them related with dashboards, I have appreciated the difficulty of building a dashboard that needs to be accurate, useful, functional, with quick answers and eye candy. Dashboards are used as a tool for the managers and executives of the organization, and need to be adapted to their requirements and likes. In this review, I’m sharing the major points to consider at the time to manage a successful dashboard project.
What is a dashboard?
There are several definitions of a dashboard. Stephen Few gave us a good example in the March 2004, article titled “Dashboard Confusion” which appeared in Intelligent Enterprise magazine:
A dashboard is a visual display of the most important information needed to achieve one or more objectives; consolidated and arranged on a single screen so the information can be monitored at a glance.
Typical examples are the simple dashboard of a car and the more complicated of an aircraft. In the case of an organization or a company, we as BI designers have to be open minded because the requirements of a dashboard project normally include:
- Multiple dashboards, each with its own objective
- Drill-down ability to analyze information
- Links to detailed reports
- Additional functionality: navigation, what-if analysis, ability to share information, internal communication, export data, print, send by e-mail, etc.
- Metadata information
Dashboard success factors you have to know
Depending on the dashboard type, you have to pay special attention to different factors. Strategic dashboards are normally used for monitoring the global company’s progress in achieving predefined goals. In this type of dashboard, users are at the top of the hierarchy, so the quality of data is very important. In addition to consolidating the information, it is important to ensure the information is correct and accurate. Nobody wants the CEO consulting the dashboard with incomplete or erroneous information. The information must be complete and valid before it is visualized on the dashboard.
Tactical dashboards are usually used for tracing the trends in relation to company’s goals and initiatives. This type of dashboard normally incorporates three types of data: detailed, summarized and historical. The users normally navigate from the first visual screen to the OLAP (On Line Analytical Process) system to analyze the information and to review detailed reports. This explains the need for a deep functional system with a quick response time, which is a challenge if the amount of data is large.
Operational dashboards are used for monitoring and analyzing the most detailed company’s activities in a given department. Normally this is related with real time or near real time, so decisions about ETL (Extract Transformation Load) process are important. Most of the operational software include their own dashboard modules. The use of this module bypasses the problem of load time, but normally with a loss of functionality. The difference of project development time is very large.
If there isn’t an indicators dictionary, you have to create it. Establishing a precise and accurate definition of metrics and indicators will facilitate the understanding of the dashboard and further promotion. End users have to understand the meaning of the dashboard if they are going to utilize it. Also, the indicators definition has to be accessible from the dashboard.
One needs to ask, is all data available? Very often, strategic data are not saved in any corporate database; instead they are saved in personal documents like spreadsheets and presentations. This presents a difficultly in the process to load the data in the dashboard.
The dashboard needs to be integrated into the organization, so it is important to consider corporative aspects: logos, colors, fonts, menu, etc. Are there similar systems? If so, use a similar look and feel.
What’s better, a very precise classic table with the exact data or the newest graphs full of colors and shapes?
The world of data visualization is changing rapidly, with the trend to use complex graphs and infographic techniques to visualize information in an impactful way. Also the concept of big data is changing our relationship with the world of information. That’s perfect for powerful presentation, but not always for a dashboard. We have to decide in each dashboard the graphic elements that better represent the business event we need to monitor. Speedometers, gauges and meters are a current trend, but they use so much space to represent only an indicator. Pie charts are good for comparison, but they are not precise. Tables are boring, but accurate. Bar and line charts are classic, but functional. A combination of areas, bars and lines on a chart is often a good choice.
The dashboard is not a static screen, since the user must interact with it. Not only print and export, but also select and filter data. Using charts as a filter should be very intuitive. Users expect tables to have drill-down functionality. The navigation has to be clear and intuitive. At any time, users need to know the level of information that they are consulting.
Manage expectations. Because dashboard tools offer more and more built-in functionality, it is common to see users waiting for the latest function they have heard or waiting for drill-down ability that you have not developed. It is important to show examples of other dashboards developed with the same tool. Building prototypes is one of the best ways to manage expectations. It has to be clear how users will interact with the dashboard.
Don’t wait until the end of the project to show the dashboard. Designing a good dashboard is not easy, no matter the experience you have. It is not possible to get it right the first time, so you have to build prototypes and quick developments to validate it with end users. They have to validate navigations, graphs, colors, fonts, data and all the important functions.
Dashboard development can have the potential to never end! This could be a great business opportunity for consulting companies but a headache for the project manager. As long as end users like the dashboard, they often want to make changes to incorporate more information and functionality. At this moment, it is important to have a limited scoped project.
As the world changes, business changes. Dashboards have to change according to business needs. Don’t forget to manage a maintenance agreement to guarantee that the dashboard will evolve according to new requirements.
A dashboard project could be an easy project to manage with few resources in a short time, or a big project that involves multiple resources with different skills: data visualization, business knowledge, database experts, technicians, consultants and managers. We need a good project definition and limited scope for making a realistic plan. To avoid failure with user’s expectations, make prototypes and rapid developments to show preliminary dashboards to the end user. If the project is large, separate it into phases for short-term results.