Business Intelligence Techniques as Marketing Tools - Essay Example

Business Intelligence Techniques Marketing has many different tools and techniques that help market researchers and marketers to plan and analyze the data and markets. Amongst the few are what we call as the Business Intelligence techniques or BI techniques. Business Intelligence (BI) techniques are the methods or ways in which one can store and use information for the business. Generally this BI covers all such tools and technologies, applications, the ways for collecting, integrating, analyzing, and presenting business data. The most interesting thing about these applications is that they use all the data that has been stored in a data warehouse, in order to report the previous business information. They also “predict future business information, including trends, threats, opportunities and patterns”. Most of these BI applications have become quite popular and are very difficult to use, thus the people who are experts in this area are quite in demand too. “Some of the currently popular enterprise level systems, which can manage information about all of the business functions and systems, are sold and implemented by Oracle, SAP, IBM, and Hewlett Packard (HP)”. Companies often need in-house experts in these systems to assist with the implementation and the on-going use of these systems, which are quite complex! The approaches for the BI can be grouped in to three areas which are as follows: 1. Standard statistical methods for quantitative data (forecasting, predictive models, decision trees, neural nets etc), 2. Semantic analysis methods (LSI, l.SA) of textual data and 3. Geographic Information Systems for spatial data. To further the discussions we will first look into all three groups individually and then compare and contrast them and also explain them in detail. Plus with the help of case studies we will also be able to shed more light onto the use of these approaches. We will also be able to use the discussion to show the strengths and weaknesses of each of the three analysis techniques above. Comparison of the different groups in the BI techniques The first group is the group of different qualitative methods used in Statistics. These statistical techniques are used widely in order to analyze and explain ‘survey datasets’. These type of analyses are very helpful in making the business owners understand the results and the interpretation of the data or information collected. In this paragraph we will discuss and evaluate some of these standard techniques and will see how these techniques can be used to investigate different patterns of data and information that is ciollected at various occasions for different purposes. Each time the data set and the collected data will be different as the purpose of the business differs. There are three standard techniques discussed in the following paragraphs namely the Forecasting method, decision tree and predictive modeling. Forecasting in other words may also be known as estimation, and is the process that is used in unknown situations. It can refer to the estimation of ‘time series, cross-sectional or longitudinal data’. The use of this method can differ between the areas of function: one of the main and salient feature for forecasting is the risk and uncertainty. ‘Forecasting is used in the practice of Customer Demand Planning in everyday business forecasting for manufacturing companies’. This method of planning for demand is sometimes referred to as ‘supply chain forecasting’ and it covers both ‘statistical forecasting and a consensus processes’. Most commonly forecasting is used in the discussion of ‘time-series data’. It is a very common and easy way to estimate the future happenings in the business although, as suggested they have risk factors involved as not always the forecasted picture is accurate and the companies have to be ready to face the consequence in cases where forecasting has failed. Our second method from this group is the Decision Tree or tree diagram. The decision tree is a support tool for business decisions that make use of a ‘tree-like graph or model’ for different decisions and their possible outcomes or results, including ‘chance event outcomes, resource costs, and utility’. More than often this method is used in areas of operations research, particularly in the decision analysis, in order to spot a most likely plan or strategy in order to reach a goal. Decision trees are also used as a descriptive means for ‘calculating conditional probabilities’. The third method of this first group of standard statistical techniques is the ‘Predictive modelling’. It is a method which makes use of a model, that is either created or chosen, for trying to best predict the probability of an outcome. Mostly, the model that is selected is simply on the basis of ‘detection theory’ in order to try at guessing the probability of a ‘signal given a set amount of input data, for example given an email determining how likely that it is spam’. Like the forecasting method this again is one of the risky methods are the reliability of this method is poor. However, like the forecasting method this too is used because of its simplicity of conduction. To further the comparison amongst the three given groups of BI tools we will now discuss the second group that covers the Semantic Analysis Methods especially including the Latent Semantic Indexing (LSI). LSI is an ‘indexing and retrieval method’ which makes use of the mathematical technique called ‘Singular Value Decomposition (SVD)’ in order to recognize the models in the associations amongst the ‘terms and concepts’ enclosed in a formless collection of text. The basic concept behind LSI is the simple rule that words which are generally used in the similar context tend to hold similar meanings. One of the main features of LSI is the ability to take out the theoretical substance from a body of text by building up relations between all those terms that take place in the same context. It is also an application of ‘Correspondence Analysis’, a ‘multivariate statistical technique’ that was developed by Jean-Paul Benzécri in the early 1970s, to a ‘Contingency Table’ that was built from word counts in documents. The salient feature of LSI is that it overcomes two of the most strict rules of ‘Boolean keyword queries’ that is the different words holding the same meanings in other term referred to as synonyms and secondly the words that hold more than one meaning or polysemy. They often cause mismatches in the language used by the authors of various documents and the users of ‘information retrieval systems’. In this case more than often the ‘Boolean keyword queries’ return irrelevant results and tend to miss information that is more relevant. The method of Semantic analysis, LSI, is also used to perform programmed document classification. Many experiments that have been conducted show that there is a correlation between the way LSI and humans process and categorize text. The term ‘Document categorization’ means to assign the documents to one or more allotted of defined categories that are generally based on the similarity of the concept in the content of such categories. ‘LSI uses example documents to establish the conceptual basis for each category’. During the process of categorization, the ‘concepts contained in the documents being categorized are compared to the concepts contained in the example items, and a category (or categories) is assigned to the documents based on the similarities between the concepts they contain and the concepts that are contained in the example documents’. The last group of the BI is the geographical survey information or the GIS. It is a technological field that includes geographical features with tabular data in order to map and analyze or assess the real-world problems.  The main feature of this technology is the word Geography. This would mean that the data or at least some part of the data is related to space. In other words it uses the data that is somehow connected to locations on the earth. Combined with this geographical data is normally the data provided in tables which is also called the attribute data.  ‘Attribute data generally defined as additional information about each of the spatial features’.  To understand this method we can take the example of schools. The definite position of the schools is based on the spatial data.  The attribute data would be the combination of additional data such as the school name, level of education taught and student capacity.  ‘It is the partnership of these two data types that enables GIS to be such an effective problem solving tool through spatial analysis.’ This method is used for many different purposes especially the ones that relate to the geographical placement of business etc. coming down to the very basic level for GIS we can say that it is sort of computerized mapping or computer cartography. ‘The real power in GIS is through using spatial and statistical methods to analyze attribute and geographic information.  The end result of the analysis can be derivative information, interpolated information or prioritized information.’ Case study comparison of the BI techniques used We now move to the second part of the assignment that will help us understand the methods or groups which have been described above. For this three case studies have been selected each showing the use of one of the three groups BI techniques discussed. The first case study is titled “A Highly Successful Predictive Modelling Marketing Campaign Slashes Home Mortgage Churn”. It involves the use of the basic or standard statistical methods. The case study is about a bank which was becoming “increasingly concerned about the number of high value mortgage customers moving to competitors”. The strategy decided and agreed upon was to persuade the customers using mortgage to continue with the bank. They did that by stressing on the substitute choices and ways they could save money on their mortgages. Calling all the mortgage customers, decidedly was too expensive and would involve too many resources.  Thus it was decided by the marketing team that they would only focus on the “high value customers” on the brink of leaving and Sysware was asked to help by carrying out a “predictive modelling exercise” in order to recognize these customers.  The case further goes to explain how the company used its system for identifying the customers on the basis of age, city, income and profession. All the information or data that they collected was entered into the ‘SAS’ prognostic modelling tool. This tool used “neural networking techniques” in order to construct a model to forecast the churn behaviour and to remove all those factors which had little or no influence. After a number of iterations that used data investigation and testing and errors, Sysware cut off the more important factors. The main idea was to reduce the number of reasons given in the model to only ten. With this method the random swings could be prevented and it would make the achievement of the best result much easier. As we have already discussed the first group consisted of basic or standard statistical methods like forecasting, decision tree etc. neural nets is also one of those methods and it has been used in the case. The case opened with the problem identified that the bank was losing its customers to competitors. The BI technique used in this case was very simple as it only aimed at allowing the bank to understand why there was a shift in the preference of the customers. For this purpose the techniques used prove to be quite sufficient and they result in what was the expected. The second case study involves the use of the Semantic LSI. The case is about “Lending Services Firm Aligns Application Lifecycle Management with Business Goals“. The case opens to show how LSI that has been designed to help in the data evaluaton functions and the banks using it benefit from it. It explains the procedure of LSI and the main features. Throughout the case study the authors have explained how after several changes and newed additions the final face of LSI was achieved. The whole study shows progress in this area and as it is quoted in the case itself that ‘LSI is now able to produce high-quality builds that accurately reflect project teams’ latest revisions. The version control in Visual Studio Team System 2008 Team Foundation Server facilitates regular integration of software code. “All the problems we had with branching in the previous system went away with Visual Studio Team System 2008 Team Foundation Server,” says Souther.’ With the help of Microsoft and its tools like Visual Studio LSI can now act in response to audit requests without difficulty, which not only saves time but helps to avoid audit exceptions. “We recently had our SAS 70 auditor come in to ask about the history of particular changes,” says Souther. “Because we had not yet migrated that project to Visual Studio Team System 2008 Team Foundation Server, the project manager needed to compile information from three sources to satisfy the request. With Visual Studio Team System Team Foundation Server, I can meet an auditor’s request simply by printing the entire history of a change.” This case study reflects on the efforts of one of the more complicated yet powerful techniques of BI and demonstrates the power of computers. It helps us to understand how technology can provide beneficial results where businesses are concerned. The third case is about using the Whatif? program for the GIS technique. It explains the use of this software for the development plan in an area near Hervey Bay shire in Australia. The case explains how computerized methods are aiding in giving a complete picture as to the development and designing of communities and to point out the most feasible decision. What if? Is a PSS software that in addition with the GIS can bring out the complete report for the development of any such community like the one discussed in the case. ‘The What if?. planning support system was used to evaluate two different outcomes for future urban growth in Hervey Bay. As its name suggests, What if? does not attempt to predict future conditions exactly. Instead, it is an explicitly policy-oriented plarming tool for considering what would happen if policy choices are made and assumptions concerning the future prove to be correct. For example, policy choices to be considered in this study include a laissez taire policy to urban growth and a policy of regulated growth. Assumptions for the future included alternative population and employment projections.’ (Using the What If?. GIS-based Planning Support System for Sustainable Development Planning) The case study ends by presenting a conclusion which states that GIS powered softwares like Whatif? Not only help the city developers but also enable general public to make their choice by presenting to them the clear picture of the future of any such decision related to investment in property or estate. These computerized softwares have eased out paths for people and business men alike. The GIS technique for the BI did benefit the people who were using it in order to evaluate the best option for the development plan of the Hervey Bay. The case had initially began with the problem that the community planners needed assistance where the decision was waiting and the opted software gave them two options to choose from. The mapping and the statistical analysis of the area and its chances of future growth were shown through the software What if? It did not predict anything simply mapped out the feasible options so that the developers could decide which one to opt for. ‘What if?. allowed the Council planners to visually compare different urban growth scenarios for Hervey Bay in the year 2021’. Thus this was one of the successful techniques for the Business Intelligence methods that have been discussed in the beginning. Resources http://ideas.repec.org/p/ags/provbp/15613.html http://gislounge.com/what-is-gis/ http://jobsearchtech.about.com/od/techcareersskills/g/BI.htm Case Study: A Highly Successful Predictive Modelling Marketing Campaign Slashes Home Mortgage Churn http://www.syswaregroup.com/About-Business-Intelligence/Solutions/Predictive+Modelling+Case+Study.html Read More