With Business Intelligence Suites attaching to more and more disparate sources of information within a corporate environment, and the processing power and availability of analytical systems within these Suites, are the three classic disciplines of Corporate Performance Management (CPM), Business Intelligence (BI) and Analytics beginning to merge into one seamless subject?
Corporate Performance Management
CPM used to be the prime driver behind improvements in an organizations performance within the market place. Utilizing methods incorporating the Balanced Scorecard from Kaplan, Six Sigma or Lean the back office staff would analyze the company business processes and seek to improve their performance by analyzing the available metrics of past performance, strategic planning, forecasting and workflow reporting. It would utilize both financial and operational performance results to provide an indication of success or failure against particular goal orientated targets, which would point to the reasons for that result.
The typical strategy for CPM would involve:
- Selection of goals,
- Consolidation of measurement information relevant to an organizations' progress against these goals, and
- Interventions made by managers in light of this information with a view to improving future performance against these goals.
Although presented here sequentially, typically all three activities will run concurrently.
BI is the process of providing for better decisions within the corporate or business area by utilizing processes, people and related data tools and methodologies. It provides both historical and current views of a company's performance by utilizing data found in relational databases, warehouses and data marts to organize historical and current information. It differs from CPM in that it is not driven primarily by a structured order of analysis against particular targets but allows the companies analysts to produce reports that inform the organization of tactical trending and opportunities. Recent developments have allowed for greater flexibilities in the data sources integrated into the BI process and this along with access to real time or near real time operation data in Point of Sale and Customer Relationship Management systems has led to the provision of 'Operational BI'.
Solving business problems utilizing statistics along with computer technology has long been known as 'Analysis or Analytics'. Originally undertaken by mathematicians utilizing formulas against company records without the assistance of computers or software these functions are now carried out by algorithms being run using data mining against large databases to extract a useful group of properties from the available data. This data is then utilized with predictive and trend models to allow the company to form a future strategy. It has however always been reliant on the quality of the data that the algorithms have to work with.
In my experience of delivering leading edge BI systems to corporate environments it can be seen that most organizations have already moved their focus from the traditional methodology of utilizing CPM to guide their corporate decision to some form of basic provision of BI against either near real or real time operational figures to provide current insight into current performance.
One of my most recent projects was to enable the service quality team of a major retail bank to assist their branch managers to show where benefits could be gained utilizing feedback provided within customer phone questionnaire results, utilizing a base OLAP database based on the month by month results of telephone surveys along with telephone and accounts data. Utilizing this base data a predictive algorithm based on survey results versus account closures over time periods that could be amended by the user allowed the central team to predict how customer satisfaction would drive account retention and overall profit trend at a branch level. This data was then embedded into results dashboards, which were presented at all levels of the company. This data led to a dramatic change in the companies' service quality rating within the industry, which conversely led to an increase in account take up and therefore profit.
This shift of emphasis from management of past performance and setting of future targets over a period of a year or more to the fast-paced reactionary management currently seen has helped to drive the requirements of the current BI Stack of applications in the market place. The ability of many of the major BI Vendors (Cognos, SAP, Microstrategy, Targit, Business Objects, IBM) to be able to set KPIs and provide a balanced scorecard in the form of dashboards based on near real or real time operational data has seen a merge of the traditional methods of CPM into operational BI as business leaders realize the benefits of the ability of these new methods.
Most Business Analysts (BAs) and managers that I speak to ask about the possibility of employing some form of analytics within their chosen solution to provide a view of historical data and trending results to enable them to predict future business performance and the changes to the customer/business models. Understanding your clients' requirements concerning their customers likes and dislikes and their propensities to either view a particular web site or buy a particular product can be used to provide better service and product offerings and therefore should be discussed at an early stage of the BI implementation. SAS have started to promote their analytics tools as part of their overall BI solution and this along with offerings from other vendors is driving the requirement for analytics to be incorporated in most corporations BI planning for the future.
Until recently, I have built and installed algorithms into bespoke solutions that I have rolled out to give the company a similar functionality to that as Clementine however, this has meant that the BI solution has been more complex than some on offer in the market but it has given the company what they wanted. However writing and implementing complex SQL Stored procedures to provide mathematical analysis of current data is at best time consuming and also not best practice as it means that the customer will require support to change the solution to look at different facts and dimensions within the algorithm if required. With SAP's new analytical tools using Clementine as its baseline and IBM purchasing outright the company (SPSS) it is obvious that it will not be long before most, if not all, BI applications will require at least a base model for predictive analysis to compete in the market. This can be seen with Tibco's delivery of their 'Spotfire' application, which I had the chance to evaluate for a customer last week. This application is more than either a BI tool or a pure Analytics platform and performs the task exceptionally well. In testing, it allowed for the BAs in the company to ask questions of the data formed in his own language while allowing the results to be delivered quickly and efficiently to dashboards and reports while allowing the business managers to look at historical and charted data with ease.
I believe that the question asked at the beginning of the article in fact has already being answered by the industries move to a range of fully featured analytical BI Suites that provide not just the ability to chart and report on historical data but also the full range of predictive analysis that is required to support the modern day business environment.