Business Intelligence Pilots

The promise of “business intelligence” (BI) has outstripped the delivery of meaningful performance. We’ll still be cleaning up data, migrating data, trying to keep our data secure and worrying about data base platform compatibility. But things are changing. Master data management is now more than an art form and data base architectures are becoming increasingly open. But much more importantly, we’re all beginning to think less about data logistics and more about analysis.

There are significant software applications that enable all sorts of data analysis of manufacturing, customer, product and transaction processes. Of course there are major data base management vendors that provide embedded and reasonably integrated data analysis tools, and there are definitely companies that run their business from analytical portals that provide insight into their operations. But there are some major problems connected with achieving this nirvana. There are still huge issues around the quality and location of data. Consequently, companies have to invest heavily in their data infrastructure and architecture to exploit BI opportunities. This investment significantly reduces the return on BI investments and sometimes even challenges the overall cost-benefit of BI. When “benefits realization” gets muddled in the executive suite, technology investments become vulnerable.

How many BI projects have failed over the past ten years? Like CRM applications, BI projects have often met resistance, cost far more than expected and failed to generate any significant business impact. Is this because BI technology is bad? It’s much more likely that the immaturity of our data environments is at the root of BI “failures” than the inherent weakness of either the BI concept or generally available BI tools.

The real impact of BI is dependent upon several maturing streams of innovation. Let’s first describe what the perfect picture looks like in, say, 2010 or 2015.

By that time business will be largely automated with a variety of if-then triggers throughout operations. Information will not be inspected by humans as much as it’s assessed by software. Business rules will drive most business processes – rules that are manually changed but much more frequently triggered by the same if-then rules that will together automate key processes and transactions. BI matures when there’s real-time insight into what’s happening, embedded judgments (rules) about what’s good and bad about what’s happening, and the automated and quasi-automated ability to do something about what’s happening.

Ultimately, all of this is an optimization equation with descriptive data feeding explanatory data which, in turn, feeds prescriptive data. For BI to be truly effective this loop must be closed – and continuous.

The first thing companies should do today is build an introspective architecture capable of providing wide and deep insight into how they operate, how they make money, how they lose money and how to ask what-if questions about the impact of alternative processes.

There are a variety of approaches here. The emerging business process management (BPM) suites provide one mapping approach. The business rules management (BRM) crowd offers also insight into key processes. Even simpler process mapping tools – like those based on systems dynamics methods – can do the job. But the key investment principle is integration so it makes sense to invest in one of the more robust mainstream process modeling suites rather than a one-off tool that may or may not be around in a few years.

Companies should inspect their data base platforms and architectures. If a company has multiple disparate data sets and non-standardized data base platforms they are ill-positioned to exploit process mapping or the full range of existing or emerging BI tools. More importantly, their ability to achieve real-time optimization will be threatened. Given the consolidation in the data base sector, it’s also wise to stay main stream with solutions in the Oracle, IBM or Microsoft direction.

Companies should also be modeling the descriptive and prescriptive questions and answers they’d like their BI environment to handle. Such Q&A exercises constitute BI requirements gathering and should position companies to think about BI today and BI tomorrow, then BI will aspire to real-time optimization of business processes and overall business models. The models should yield a set of rules about what-should-happen-when and rules that represent optimal corporate performance. Today, these rules are captured in rules engines that are usually integrated in a limited way but tomorrow will be fully integrated and dynamic. The key to modeling today is to apply white board thinking to what optimization is as opposed to thinking constrained by what we know the current capabilities look like.

Companies should be piloting existing BI tools for their ability to support descriptive and prescriptive questions and answers. Huge investments are probably not advisable at this point for most companies, especially if they are still struggling with platform standardization and larger data quality, security, access issues. But as these issues disappear, companies should strive to identify the processes, methods and tools most likely to satisfy their expected BI requirements which, again, will all tilt toward real-time optimization.

We assist in the definition, conduct and assessment of low-level and high-level BI pilots.