Copyright(C)2009 NTT DATA Corporation A Business Process-IT Alignment Method for Business Intelligence Jun Sekine, Takashi Suenaga, Junko Yano, Kei-ichiro Nakagawa, and Shu-ichiro Yamamoto NTT Data Corporation
Copyright(C)2009 NTT DATA Corporation 2 Introduction The concept of business intelligence (BI) is to provide information for achieving business goals. The BI systems require business process -IT alignment in order to provide functionality useful for achieving the business goals.
Copyright(C)2009 NTT DATA Corporation 3 Issues (1) BI systems vary depending on the maturity of target enterprises. Ex. 1: Visualization of KPIs and related data using OLAP tools Ex. 2: Validation of hypotheses for achieving business goals using data mining methods KPI: Key Performance Indicator OLAP: OnLine Analytical Processing ⇒ We need a business process-IT alignment method for BI covering all the variations.
Copyright(C)2009 NTT DATA Corporation 4 What Is A Hypothesis? A hypothesis is a written statement that may achieve business goals (KPIs corresponding to effects). comprised of at least one causal factor (KPI corresponding to causes) affecting the business goals. Ex. Outsourcing of parts of work of projects improves profits. Causal Factor Business Goal Actions Control
Copyright(C)2009 NTT DATA Corporation 5 Issues (2) BI analysts need to convince enterprise management that their proposed hypotheses and actions are comprehensive and rational. ⇒ We need a kind of framework used to clarify and organize the hypotheses and/or the actions.
Copyright(C)2009 NTT DATA Corporation 6 Our Approach A business process-IT alignment method for BI with two phase business processes for BI defining information that should be managed by the BI systems covering four different levels of BI systems, with a modeling process of the framework clarifying hypotheses and actions helping enterprise management understand their target domains.
Copyright(C)2009 NTT DATA Corporation 7 Business Processes for BI ~ Phase 1 Phase 1 Extracts and checks the validity of hypotheses for achieving business goals. Phase 1 Extracts and checks the validity of hypotheses for achieving business goals. 1.Define business goals 2. Create model 3.Extract hypotheses 4.Choose valid hypotheses 5.Choose achievable hypotheses 6.List & validate actions 7.Execute actions 8.Monitor actions 10.Improve hypotheses 9.Improve actions Phase 2 Clarifies the actions needed to implement the hypotheses for achieving goals. Phase 2 Clarifies the actions needed to implement the hypotheses for achieving goals.
Copyright(C)2009 NTT DATA Corporation 8 2 Ways for Extracting Hypotheses (1) Extract hypotheses through interviews. Use a model for mapping the hypotheses and checking their completeness. 2. Create model 3.Extract hypotheses Planning & prepa- ration Sales Service ordering Starting service Charging Payment Business Tasks Decrease the number of complaints of an electronic apliance company it receives from service agents Types of complaints were categorized by the business tasks of selling products at service agents Business Goal Model
Copyright(C)2009 NTT DATA Corporation 9 2 Ways for Extracting Hypotheses(2) Models are sometimes created auto- matically using data mining methods. Ex. 1: Clustering customers based on their purchase history. Ex. 2: Viewpoints of classifying profits, such as the locations of branches or the categories of products (These are usually called “dimensions” in OLAP tools.) 2. Create model 3.Extract hypotheses
Copyright(C)2009 NTT DATA Corporation 10 Business Processes for BI ~ Phase 1 Phase 1 Extracts and checks the validity of hypotheses for achieving business goals. Phase 1 Extracts and checks the validity of hypotheses for achieving business goals. 1.Define business goals 2. Create model 3.Extract hypotheses 4.Choose valid hypotheses 5.Choose achievable hypotheses 6.List & validate actions 7.Execute actions 8.Monitor actions 10.Improve hypotheses 9.Improve actions
Copyright(C)2009 NTT DATA Corporation 11 Business Processes for BI ~ Phase 2 1.Define business goals 2. Create model 3.Extract hypotheses 4.Choose valid hypotheses 5.Choose achievable hypotheses 6.List & validate actions 7.Execute actions 8.Monitor actions 10.Improve hypotheses 9.Improve actions Phase 2 Clarifies the actions needed to implement the hypotheses for achieving goals. Phase 2 Clarifies the actions needed to implement the hypotheses for achieving goals.
Copyright(C)2009 NTT DATA Corporation 12 Choose Achievable Hypotheses Ex. The profits of franchise shops can be increased by locating them near stations. Actions Causal Factor (Distance from neaby station) Causal Factor (Distance from neaby station) Business Goal X Moving existing shops is impossible.
Copyright(C)2009 NTT DATA Corporation 13 Distribution of Data Analysis Cases
Copyright(C)2009 NTT DATA Corporation 14 Maturity Levels of BI Creation of decision support info. Mgmt. of info. Mgmt. of actions (Phase2) Mgmt. of KPIs (Phase1) Level 1: Visualization Level 1: Visualization Level 2: Data mining Level 2: Data mining Level 3: Validation of actions Level 3: Validation of actions Level 4: Service execution Level 4: Service execution Scope Functionality Visualize KPIs, models. Validate hypotheses. Validate the effectiveness of actions. Create models or hypotheses using data mining. Create new services using statistical analysis, etc.
Copyright(C)2009 NTT DATA Corporation 15 Models Frameworks for categorizing and organizing hypotheses and actions, and provide accountability for enterprise management. Three types of models: Type 1: Categorization of business objects Type 2: Description of relationships among business objects Type 3: Description of business tasks
Copyright(C)2009 NTT DATA Corporation 16 Type 1: Categorization of Business Objects Categorization of business objects (BOs) that are the target of improvement. Shops in the suburbs Shops in downtown Ex. 1 Categorization of shops. Increase the sales of shops Target Medical expenses Time Hypertension Atherosclerosis Cerebral infarction Ex. 2 Transition processes between diseases. Decrease medical expenses Target
Copyright(C)2009 NTT DATA Corporation 17 Type 2: Description of Relationships among BOs Description of business goals, BOs consisting of business environments, and the relationships among them. Employee Project Member Organization Employee Satisfaction Assign Education (Skill) Outsourcing Partner Profits (sales, costs..) Member Increase the profit of each project Hypotheses A)High degree of employee satisfaction leads to large profits. B)A certain number of skilled managers is necessary. C)Proper outsourcing of part of the work is necessary.
Copyright(C)2009 NTT DATA Corporation 18 Type 3: Description of Business Tasks Description of business tasks that enterprise management would like to improve. Planning & prepa- ration Sales Service ordering Starting service Charging Payment Decrease the number of complaints it receives from service agents
Copyright(C)2009 NTT DATA Corporation 19 Conclusions A business process-IT alignment method for BI with two-phase business processes, covering four different levels of BI systems. with a modeling process, helping enterprise management understand their target domains.
Copyright(C)2009 NTT DATA Corporation 20 Further Issues It is not easy to reach levels 3 or 4. We should start from levels 1 or 2, and then proceed to the next levels. An education program The types of models that are used in different situations are yet to be investigated. A catalog of modeling methods