Last week I found myself wondering why most Business Intelligence (BI) projects fail to deliver the expected results. I have been stumped by the number of well-funded BI implementations that are seen gathering dust at firms. It’s probably why many people think implementing a BI solution is a long, drawn out, resource intensive project that is doomed to go over budget. But they don’t need to be. In this and two follow on posts, I will explain the biases many corporations have towards BI implementations as well as the steps you can take to ensure a successful implementation.
Let me begin by pointing out that BI solutions are substantially different to core IT systems and there is a significant difference between implementing the two. Most companies fail to differentiate this resulting in a sub-optimal BI implementation. Successfully implementing a BI solution requires a paradigm shift from that of implementing a core IT system. And the shift is hard to make due to entrenched mindsets and processes at companies and their software vendors.
First, let’s talk about core IT systems. Every company understands the importance of ERP, CRM, accounting, and inventory management systems. These are transactional applications that enable businesses to carry out their day-to-day operations and ensure efficient transfer of data across the organisation. Therefore, these projects are usually large-scale, expensive, and involve most of the employees in the organisation. I consider these systems akin to the nervous system in humans.
A BI solution on the other hand is equivalent to the brain. The two systems should work together, processing information and generating (mostly) rational responses to stimuli. In a corporate environment, BI systems are designed for management to make better decisions and therefore by definition have fewer users of the system. Whilst most executives readily acknowledge the need for information based decision making, the vast majority make do with reports churned out by the transactional systems. However, transactional systems usually do not talk to each other and this makes it extremely difficult to conduct cross-functional analysis.
A BI solution has a singular objective – enable management to make better decisions, faster. In order to achieve this objective, the solution should be able to ingest multiple data sources and conduct relevant analysis that finds the inter-relationships between datasets. This should form the basis for hypothesis testing and scenario analysis that will help executives understand the implications of pulling different levers and the impact it has on their KPIs.
The differences in the organizational reach and utility of these two different types of systems means that the approach to implementing them should be different too. However, more often than not, BI implementations are done in the same way as core-system implementations.
So what does it take to successfully implement a BI solution in your organisation? How can you ensure there is maximum ROI on your efforts? What are the different phases and what should you focus on in each phase? I will touch upon these in the following post.