How predictive analytics enables enterprises to stay ahead of market forces

The credo ?You can?t improve what you don?t measure? underscores why decision makers in an enterprise require timely, accurate and comprehensive information about business performance. Over the past decade, many different business performance management BPM- software products have emerged to address the challenge of putting actionable information on every manager?s desktop.

Most of these BPM solutions do a fair to excellent job of collecting, consolidating, validating and reporting enterprise data from a variety of sources in order to give decision makers a clearer overall view of how the business is performing. However, the crucial task of analyzing what this data signifies about current performance within the business ?and, more importantly, where it?s headed in the future? is still cumbersome. A manager who wants to understand a variance in business results i.e., the gap between expected versus actual performance- must first identify the key performance indicator that is out of tolerance and then manually drill down within the historic data for further detail. This difficult, after the fact and time-consuming search for the reasons and root causes of a variance will not only leave managers with little time for actual decision making, but will also make it harder to take action in a proactive manner. Today?s chief executives demand faster answers to the deeper questions ?where their business is doing well, where the problems lie, and what should be done to correct those problems? on a daily basis. What?s more, the Sarbanes-Oxley Act and similar legislation is requiring enterprises to improve their decision support systems in order to reduce reporting times for 10Q?s, 10K?s and all material events. In short, variance and root cause analysis are critical necessities for business managers seeking to better understand business performance, quickly spot opportunities or problems, and take corrective action where necessary. What?s missing from current reporting and BPM software systems is a more proactive and automated way to understand the reasons and root causes driving performance variances - all delivered in a relevant context. Providing automated root cause discovery and analysis in context, along with the ability to predict future variances, is the logical evolution for BPM software. Rather than laying the burden of analyzing variances on a manager who may not have the time or technology skills to properly investigate the data, what?s needed is a means of automating the discovery and prediction of variances and delivering that information in the appropriate context for each user. This concept of employing predictive analytics -to understand in real-time why a variance occurs and what to do about it moving forward- represents the next evolutionary step beyond current BPM software offerings. Predictive analytics within a unified BPM framework ?one that consolidates data from all of a company?s important business processes e.g., planning, budgeting, forecasting, etc.- within a single application platform? should provide the following core capabilities: The ability to automatically discover variances and analyze the root cause or causes in business performance data that exceed their threshold Tools that act upon the root-cause analysis results to generate predictions -- in the form of on-demand forecasts, alerts or early-warning indicators -- about other likely performance deviation scenarios and their impact on overall business performance. A means of delivering the above information to key decision makers in the appropriate format and context. These three pillars provide a foundational basis for evaluating the relative strengths of different BPM software frameworks in the area of predictive analytics. To gain a clearer picture of the specific technologies required to facilitate predictive analytics, it?s helpful to first define what?s meant by the concepts described above. When an unacceptable variance arises in any key performance indicator KPI-, business managers need to uncover the underlying reasons. The predictive analytics software should automatically and proactively bring such variances to the surface, highlighting all under-performing and over-performing KPIs on the dashboard of the BPM system. From there, the software should break down the KPI into its component parts ?such as actual versus budgeted monthly revenue ?and examine the variations among the components. As part of this variance analysis, the software also should give business managers the option to drill down from the dashboard view to a detailed examination of underlying reasons for the variance, as well as view descriptions of key transaction-level events that lie at the root of the variance. For example, clicking on the total revenue figure for a given month should show the user a breakdown of sales figures for the products that contributed most heavily to that month?s revenue shortfall, along with a look at which customer sales contracts fell through. The predictive analytics system also should be able to perform a detailed search and retrieval of any unstructured information within the BPM system that could provide further context surrounding the variance, such as a sales recap in PowerPoint or e-mails from the customer account representative. Building upon the software?s ability to discover performance variances and correlate them to changes in the KPI?s underlying sources, and recognizing how those variances are likely to affect future performance, is a key advantage of leveraging a unified BPM solution equipped with predictive analytics capability. Any change in a base-level data point, whether known or forecasted, should trigger the system to create a set of early warning signals that can be sent to the decision makers responsible for taking corrective action. Such an alert should identify the specific KPI at issue, the amount of variance, the percentage level of confidence or likelihood that this situation will result in a future positive/negative result, and the underlying analytical basis for the prediction. The value gained from automatically discovering KPI variances and predicting their impact on future performance can only be fully realized after this information reaches a decision maker who can act upon it. And before a business manager can take effective action, he or she must first understand what has been presented. Therefore, the BPM software needs to provide an appropriate context and level of predictive detail based upon each user?s role and responsibilities within the organization. Similarly, the system should be able to display the same core information in a variety of formats. For a chief operations officer who needs a quick overview of high-level KPI results, the variance analysis may best be depicted in a 360-degree radar screen by contrast, a line-of-business analyst is more likely to require in-depth charts paired with transaction-level KPI indicators. In all cases, though, the software should provide immediate click-through access from any of these formats into the same underlying levels of detail generated through the discovery and prediction phases. With predictive analytics capabilities incorporated into a unified BPM framework, decision makers can dramatically reduce the time and expense involved in moving from the discovery and reporting of a KPI variance to isolating its root causes and taking well informed actions to mitigate or prevent future problems. Armed with deeper insights into the real causes of business performance variances, in addition to the context surrounding them, managers can better predict tomorrow?s outcomes and make proactive decisions that enable the enterprise to stay ahead of competitors and market forces. Source: www.businessintelligence.coma>