Business intelligence is awash in so many products, buzzwords and technologies that it is often hard to see the forest for the trees. This is but one example of how to use the three M s when analyzing the business intelligence market.
The first and oldest category is manage. It began with COBOL-generated green-bar paper reports in the late 1960s, evolved into computer output laser disk COLD- in the early 1990s and enterprise reporting and OLAP cubes after that. When all is said and done, these are detailed reports that middle managers use to run the business: production plans and balance sheets from ERP systems, for example. Nothing fancy, but the lifeblood of any modern corporation. If any one of these standard reports doesn t turn up on time, sit back and watch the fireworks erupt. Integral to such reports is the ability to handle large amounts of detailed data and to roll it up in a standardized fashion. By dint of hands-on experience, line managers know what to look for - if a number is off, they can usually dive down and figure out the problem by looking at the reports or making a few phone calls. Monitor, the next category, evolved as the corporation wanted to distill daily operational data so it could be quickly comprehended by upper management. Mid-90s mechanisms such as dashboards, key performance indicators KPIs- and balanced scorecards fall into this category. Note that monitoring couldn t occur until vendors and enterprises had manage straightened out. Monitoring systems imbue summarized transaction data with corporate strategy. This is in contrast to managing systems, which by dint of their detailed transaction orientation, are relatively similar from company to company - every company needs an accounts receivable system or a balance sheet, for example. Monitoring is where companies display their strategic uniqueness - by deciding on a specific set of metrics within a balanced scorecard, or by designing a certain dashboard style. Put another way, monitoring is where companies decide what to track closely - and, consequently, what to ignore. If a KPI is out of range, it usually takes more than a glance at a report or some phone calls to figure out the problem. Because it is a distillation of several or even numerous metrics that may be pulled from various departments, a KPI is a complex and not always straightforward thing. Therefore, although a KPI is not a good diagnostic tool, it is a highly effective warning device - and can highlight which sections in manage need to be investigated. Both manage and monitor are reporting functions - that is, they track past history. In contrast, model, the last category, looks to the future. At times, model means predict - by leveraging algorithms, models and large amounts of data, data mining can do a good job of forecasting customer behavior, for example. But model can also mean pondering the possibilities - performing what-if exercises not to predict the future but to understand how it may evolve. These simulation exercises can serve the purpose that war games do in the military - helping personnel plan for the possible options, so that when confronted with a change or surprise, they can react coolly under pressure. In the past, such planning was performed only with financial figures, as in budgeting and planning. However, as more data becomes available, companies are modeling other parts of the business, such as how to react to a changing mix of customers over time. These two flavors of model - predict and plan - are relatively labor-intensive, compared to manage and monitor. Once set up, managing and monitoring systems can perform prodigious feats of analysis, working through terabytes of data that would take humans months if not years to comb through. In manage and monitor, the technology performs 80 to 90 percent of the work. In predict and plan, technology and humans share the work equally. Good data mining requires statisticians and a thorough understanding of the business to find the best model and get the best lift. Strategic planning also requires a thorough understanding of the business as well as the very human ability to think outside the box. Due to this relative labor intensity, model has not been adopted as fast as manage or monitor. However, that doesn t mean it doesn t have value within the BI toolkit. While the major BI vendors are moving to offer all three of these capabilities within their suites, one BI tool has had all three for years, which I think helps explain its popularity. That tool is Microsoft Excel, which because it is generic enough - read, Not optimized for any of the three but capable of all three - has been pressed into service to handle the different BI tasks at hand. After all, a mid-level manager can use Excel to put together a rudimentary reporting system a C-level executive can use it pull data together from various departments to monitor corporate health and a strategist can use Excel formulas to generate what-if scenarios. Bron: www.datawarehouse.coma>