Choosing the right business intelligence software is often a critical, but difficult, decision for businesses. Jeffrey A. Shaffer, vice president of analytics at Recovery Decision Science, outlines the key factors that should be considered before selecting a new BI platform.
Selecting business intelligence software can be a daunting task. It’s a decision that can shape the analytics and reporting capabilities of a company or organisation for years to come. In addition to the high costs of the software and implementation, there are many other considerations that companies should be aware of but which are often overlooked. This article outlines a number of important factors, beyond costs, which should be taken into account.
One of the most important considerations in the software package selection process is the functionality of the software. Gartner publishes an annual Business Intelligence Magic Quadrant and Critical Capabilities Report, which focuses heavily on functionality and outlines 15 key capabilities of business intelligence software platforms. These key capabilities include administration, security, connectivity to various data sources, workflow integration, ease of use and visual appeal. Many of these capabilities will be integral to the decision process, therefore, this report will undoubtedly be a great resource for anyone buying business intelligence software.
Most business intelligence software vendors have built data connections to most of the popular data sources. Having said that, it’s still important to consider all of the data sources in the environment and assess this connectivity. Some vendors now have native connections to things like Google Analytics, Salesforce, Google Sheets and other very specific data sources. In addition, it’s important to know how these data sources can be joined or linked together to provide the necessary detail for analysis or reporting.
Ease of use, iteration and flexibility
It’s very important to look for software that is easy to use and allows you to iterate quickly.
The Big Book of Dashboards features 28 dashboards from 17 different contributors, each covering a different business scenario. Each dashboard in the book has gone through multiple iterations. In some cases, the changes have been very minor, but in other instances major design changes have taken place.
If the iterative process requires the user to request changes back to an IT department that then has to develop and deploy each iteration, it could take several months to develop the final product instead of weeks or even days. A user should be able to bring their data into the tool, quickly create visualisations that allow them to explore their data and quickly move from visual to visual, or field to field within a visual, quickly and easily. They should be able to stay in the flow of analysis without needing the assistance of a support team or developer team to these make changes. Imagine a meeting where a dashboard is showing sales by state, but the executive would like to see that data at the county level. Will the tool support this view modification during the meeting at that very moment?
There is a learning curve on any new software. As people first learn, and continue to learn the software over the years, it is very important to have resources to support them. This goes beyond the vendor’s support team. Look for resources online, training programs, workshops and conferences. Some of the BI platforms have very strong communities of bloggers, offering “how to” posts or “tips”. This also includes user activity on forums or social media where questions can be posted and answered quickly by the community. Look for users who have built successful internal communities around the tool, for example, a Centre for Excellence. These resources can be invaluable when deploying large-scale implementations and training programs.
Look for vendors who have devoted time and effort into best practices. For example, one of the most important considerations in data visualisation is the use of colour. The three primary ways of encoding colour in data visualisation are sequential, diverging and categorical colour schemes. Experiment with these BI programs and closely examine how colour is being encoded by default. Is the program encoding a categorical comparison with a sequential color? Is this easy to change?
Static visualisation vs interactive visualisations
Find a tool that can create both static and interactive visualisations. The tool should allow the user to export a PDF or image as needed, but, more importantly, it should offer a robust solution for interactive visualisations. The tool should allow for filtering and highlighting in a simple and easy manner. It should allow for building visualisations at summary level for an overview and then allow the user to filter and see details on-demand.
Analytics capability and integration with statistical programming languages
In addition to visualising data, it is also important that the tool has strong analytical capabilities and integrates well with other statistical programming languages such as R and Python. Look for software that offers core functionality but has the capability to connect and integrate in cases where additional functionality is not native. For example, a tool might have basic forecasting functionality that is sufficient, but it may not have built-in capability to perform a market basket analysis. In this case, having the capability to integrate with R or Python will allow for many more analytical capabilities from within the tool.
Implementing a business intelligence platform is a critical decision. Choosing the wrong package can waste money and time. Business intelligence platform implementation can be transformational to a business, and having the flexibility and ease of use will allow the company to extend the development to more users outside of the IT department, and enable quick to iteration and modification, without the need to code. The end result is far superior and will take much less time – often a few weeks or even in a few days.
Jeffrey A. Shaffer is vice president of analytics at Recovery Decision Science. He is also adjunct professor at the University of Cincinnati in the Carl H. Lindner College of Business, teaching data visualisation. He has taught data visualisation and Tableau at Fortune 500 companies, top accounting firms, large non-profit organisations and government agencies.