Starting small with the big picture vision is key to BI success

Paul Morgan, managing director of ASYST Intelligence (http://www.asystint.co.za), a focused provider of strategic business intelligence and data management solutions, says that one of the main reasons why companies are not seeing significant returns on their investment in BI is because they are trying to do too much at once.

“It's a clear case of paralysis by analysis. Companies have a good idea of what BI can do for them and they have eagerly adopted the technology with big ideals and bold goals for an enterprise-wide BI strategy. But, before they even begin implementing an enterprise-wide BI strategy, they have to establish a central data store from where BI tools can extract the vital intelligence they are hoping to make more accessible across the organisation.

“The problem is that it can take years to implement an enterprise data warehouse because companies typically have a range of different systems bulging with different types of information that needs to be cleaned, transformed and loaded into the data warehouse. And, these systems, if they are not standardised, are not always easily integrated, which also serves as some explanation as to why some companies that have succeeded in implementing an enterprise data warehouse are still grappling with disseminating information throughout the organisation,” he says.

“Unfortunately, many companies also underestimate just how complicated it can really be to integrate BI tools with their existing IT systems. The result is that BI implementations are hugely complicated and end up taking years, during which time enthusiasm begins to wane because people just don't see the results or the value of BI.

“Rather than set their sights high on an enterprise-wide BI strategy, companies need to start small, with the bigger picture in mind,” he says.

Morgan says the trick is to start with integrating one or two key systems such as the sales and supply chain systems for example; build the architecture quickly, and start delivering dashboards, reports and queries as soon as possible. The parallel is with Extreme Programming (XP) in the software development industry – people need to see results quickly so expectations can be clarified and met.

“Our recommendation is to limit each project iteration to three months. If you can't produce something tangible and useful in that time then maybe you're not using the right tools.

“By starting off small, companies can focus on getting it right in one area or two areas and see results much faster. People like to see results and results drive user acceptance. Aside from ensuring buy-in, because business can immediately see the value, companies can also apply the lessons learnt in other areas as they build the enterprise data warehouse bit by bit and roll out BI across the rest of the organisation.

“Also, don't forget that we operate in a fast-paced business environment where systems and needs evolve continuously. This kind of approach allows companies to adapt their BI strategy to meet their changing strategic requirements as they go along,” explains Morgan.

Companies should anticipate reworking some of the solution to fit newly uncovered requirements as you progress.

“This doesn't mean that companies should avoid thinking about the big picture, but the modular approach helps to define exactly what the big picture will end up looking like. The key is to build for requirements now. The ‘Big Design Up Front' methodology won't ever work in large BI deployments because organisational needs change faster than large-scale solutions can be built.

“Obviously, keeping the momentum going is critical to avoiding stagnation. Parallel to one or two systems being successfully integrated and BI results being delivered, plans must be underway to start integrating other areas of the business. Companies also need to plan for some overlapping and accept that they will need to make changes as the enterprise data warehouse grows,” advises Morgan.

According to him, companies are waking up to the importance of standardising on their IT systems. This is especially important when it comes to the BI tools they select. If there are a range of different technologies that have already been selected, companies should seek to select flexible BI tools that are able to work off different platforms.

Ideally, all BI tools that sit on top of the enterprise data warehouse should be standardised. However, there are sometimes certain issues that can only be addressed with niche solutions so companies may have to make some exceptions. In these instances, it is important that everyone including the teams responsible for information management, BI and the data warehouse are involved in the application selection.

Morgan reiterates that companies should conduct a small-scale pilot project using the BI tools on a minimum of two applications before they commence with rolling-out to the rest of the business. This should involve a thorough analysis and evaluation of the solutions' appropriateness.

“It is important to conduct a proper proof of technology before deciding which BI tools to standardise on. It is advisable to commission the help of BI specialists with the appropriate skills and resources to offer guidance on solution selection, as well as provide support during implementation.

“By taking BI one step at a time and enlisting the help of qualified BI consultants to guide them along the way, companies can significantly increase their chances of BI success,” concludes Morgan.