27 May 2019
Johannesburg - "Business intelligence still has the same meaning, which is to analyse information and make it actionable. But it's the scale and scope that has changed, from business information in Excel sheets to now incorporating sources such as IOT," explains William Brandon, Manager at DBA Architect.
Modern BI has expanded quite drastically under the auspices of digital technology. The practice is familiar enough: every business generates information that is worth analysing for new insights. It may be because a branch office is underperforming: why this slump? What can be done about it? Those answers reside in analysing the business information at hand.
Past BI exercises were designed around scarcity of resources, information or both. But today's digital cultures have flipped that on its head: information is abundant and the means to analyse that data is very affordable.
"It's easy to crunch a lot of data," Brandon explains. "To humans, a hundred lines of information is a lot. But to a machine, it's almost irrelevant if there are a hundred or a million lines. It can create a much richer context than we were ever capable of before."
Yet, for all their promises, most modern BI projects fail, a staggering 70% to 80%, according to Gartner. Reasons for this are distilled to two points: a lack of communication between IT and the rest of the business, as well as asking the wrong questions.
It serves to start with the latter. Businesses can become very excited about BI's potential, but completely neglect their role in making the technology work for them. According to Brandon, BI shouldn't be approached as a technology project that will deliver good business outcomes. It is fundamentally about answering business questions, so that's where conversations ought to start: "The first step of BI is always understanding the business' needs and agreeing on which key points to focus. After that, you need to create a data map in order to understand what data is collected, where it comes from and how it flows through the business. Once you have mapped out all the relevant data you will need to identify which data you need and highlight any areas where crucial data is missing."
BI projects do not need to be overarching presences across the business. They can start at a specific place to create visibility there. Taking the earlier example of lagging branch sales, a department might strain to find satisfactory answers in the relevant data. This is a good place to deploy modern BI and thus an obvious business case for it. So, where does it go wrong?
Such an approach is often oversimplified: business minds look for KPIs and similar straightforward metrics, so the due diligence of creating the right questions and mapping data accordingly is often neglected. It touches on the other problem spotted by Gartner: a lack of communication between IT and the rest of the business. Business departments might treat BI as another IT project from which they only want outcomes.
But, unless all the right stakeholders are around the table, BI's failure is all but assured. IT can do all it might muster, yet the understanding, appreciation and modelling of data reside with the relevant business department. It's the department, not IT, that determines the culture and integrity of information: "BI reporting can only be as detailed as the raw data it's based on, so the biggest barrier in BI is data integrity. You would be surprised at how many companies use numbers that a clerk writes on the back of an envelope."
Creating BI from a broad scope is challenging, hence it's better to start such projects on more localised level, such as for specific departments. BI developers prefer to work with key departments and their data, aiming at specific questions.
The good news is that today's BI platforms can start with a small footprint: proofs of concept are often the genesis of larger BI presences. It's also a boon that BI is not an input technology: it does not feed data back into the business systems, but presents answers for leaders to mull. This vastly reduces the complexities that are associated with systems integration.
Instead, the face of BI is the dashboard: a colourful and malleable interface presenting data as graphs and other visual interpretations. The point is to deliver actionable information to relevant users as they need it. Market-leading BI platforms allow for the easy customisation of dashboards by business users, whereas poor user interfaces often contribute to failed projects.
This lack of meddling with other business systems also makes BI projects ideal springboards for pursuing a larger digital transformation agenda, as they involve many of digital's principles without meddling with current technology investments. One can start with existing reports and data, then experiment with advancing features such as automated capture of information.
"There is little downside to modern BI platforms if they are done with the right reasons and attitude," Brandon concludes. "The misconceptions around cost and complexity are misplaced. If the data is ready, it can take as little as a few days to get a POC ready. But, it will only be successful if the parts of the business that want the insight approach BI correctly."