A Process-Centric Approach to Business Intelligence
Vendors and analysts often talk about the need to make business intelligence (BI) process centric. Despite this ongoing discussion, there is no clear definition of process-centric BI and little has been written about its benefits compared with a data-centric approach. This situation gives me the opportunity to hopefully clarify matters by offering my perspective on this topic.
When designing and deploying IT applications, operational business transaction application developers have a different mindset from BI developers. Business transaction developers define applications in terms of the business processes they manage, whereas BI developers view applications in terms of the data they handle and maintain. To understand why these different viewpoints exist and how they affect process-centric BI application development, we need to explore four types of IT applications and processing that support business operations – operational business transaction processing, collaborative processing, BI processing, and business planning. Master data management is a fifth type of application processing, but for simplicity we will ignore this type in this discussion.
Understanding Business Transaction Processing
The main objective of business transaction applications is to support the day to day business operations of a company. When processing customer orders for computers, for example, these applications validate and enter orders, schedule the assembly and testing of the computers, handle the shipping of the orders, and bill clients. To support this order processing other applications track inventory levels, order parts, and so forth.
Business transaction processing handle the many interrelated business processes that work together to meet the operational needs and goals of the business. Each process consists of one or more activities that perform specific operational tasks that need to be carried out. The way these business processes and activities interact with each other can be described in business transaction workflows. These workflows may be documented on paper, in spreadsheets, or using workflow tools.
As the IT industry has evolved and offered better software solutions, more and more business operations have become automated. This automation was initially done using monolithic applications that exchanged information using batch files. Many of these applications have now been replaced by online applications that communicate using distributed computing technologies. Early distributed computing approaches were time consuming to implement and involved proprietary techniques and technologies. Organizations are now solving this problem by moving toward interconnecting business transaction applications using a service-oriented architecture (SOA). An enterprise service bus (ESB) is sometimes used to manage the SOA environment and to connect legacy applications to the SOA environment.
Business transactions and their associated workflows involve application to application (or service to service) communication with possibly some user interaction. When user involvement is required, the workflow interacts collaboratively with the appropriate users as required. There are many business process management tools on the market that handle process workflow definition, automation, and integration. A large proportion of this software comes from infrastructure vendors such as BEA, IBM, Oracle, Microsoft, and SAP.
The Role of Collaborative Processing
Collaborative processing supports the manual and semi-automated business processes that exist in organizations. Order processing for business computers can be fully automated when orders are entered via the Internet or corporate extranet, but this is not the case when a customer sends in an order using a fax, e-mail or the postal system. These orders have to be entered manually and the appropriate paper documentation imaged in case it is required in the future. Other examples of semi-automated or manual processes include claims processing, loan applications, customer service calls, and purchase order creation and approval.
Collaborative processing tasks and relationships can be defined and documented in a collaborative workflow. These workflows are oriented more toward people-related tasks compared with business transaction workflows. Several vendors offer software solutions that aid in the documentation of these workflows. These tools also help with automating collaborative tasks and managing the unstructured data associated with them. The infrastructure vendors mentioned above, and enterprise content management and collaborative vendors such as Adobe, EMC, Filenet (recently acquired by IBM), and OpenText provide these types of tools.
Until recently, business transaction workflows and collaborative workflows have remained separate, but with the industry moving toward workflow standards (such as BPEL and BPMN) and an SOA environment these workflows are being brought together. This effort is coming from a number of different vendors, but companies like Adobe, IBM and Microsoft are putting significant resources into providing shared workflows, common forms handling technology, and document management services.
Enter Business Intelligence
So far then we can see how business transaction and collaborative applications work together to support daily business operations. The third type of processing is business intelligence, which has the objective of monitoring, reporting on, and analyzing business operations. BI is not involved in running business operations, but in looking for ways of making those operations more efficient and more effective.
There are three types of BI – strategic BI, tactical BI, and operational BI. Initially, most BI applications were developed for business analysts and experts whose daily job involves accessing and analyzing data. These BI applications were tactical in nature and were targeted at making short term business decisions involving marketing campaigns, the budgeting process, and so forth. More recently, BI has been extended to support executives, and senior and line-of-business managers (LOB). Here, the BI is used to support long-term corporate goals and objectives that we see in annual company reports like reducing costs and increasing revenues. These long-term goals usually drive the short-term initiatives measured by tactical BI applications.
Strategic and tactical BI applications provide useful information or measurements about business performance, but by themselves they do little to help business users manage performance. For users to manage performance, the BI needs to be put into a business context. This is done by tying actual business performance data to planned performance data. This requires BI applications to be integrated with business planning processes managed by spreadsheets, standalone planning tools, and ERP business applications. Tying the two types of data together produces so-called actionable BI that can be delivered to business users through business planning dashboards. When actual performance and planned performance are not aligned, the responsible business owner can then take action to close the performance gap.
Strategic and tactical BI is primarily data centric. This is because the planning objectives, goals, initiatives, targets and thresholds (such as sales revenues and product marketing costs) that BI is being used to monitor and measure are also data centric. If it is necessary to tie the BI results back to specific operational business processes then this is typically done manually by business users.
To date, vendors have paid limited attention to the role of process workflows in strategic and tactical BI, or in linking this BI back to operational business processes. This, however, is starting to change. Some vendors are now providing BI workflows with their products. The concept here is that the analyses and steps an experienced business user employs to make a decision, or take an action, can be captured in a workflow and replayed at a later time by other, possibly less experienced, users. These workflows guide business users through the use of related reports and analyses and assist them in finding the information they need to make informed decisions. Unfortunately, most of these guided BI workflow techniques are proprietary and do not exploit the more standardized workflow approaches used in business transaction and collaborative processing.
In terms of linking BI to operational business processes most of the effort has been focused on strategic BI. Methodologies such as strategy maps are being used to understand the cause and effect of specific business initiatives on each other. An example would be to compare and relate demand-driven initiatives with supply chain goals, i.e., to relate supply to demand.
What About Operational BI?
Operational BI is used to manage and optimized daily business operations, and the concepts and techniques discussed above for tactical BI apply equally to operational BI. The only real difference between tactical and operational BI lies in the granularity of the data being analyzed and the frequency at which it is being captured, analyzed, and reported.
A process-centric approach to operational BI is fertile ground for helping manage daily and intra-day business operations. A suitable analogy here is a manufacturing line where sensors are used to gather data at various points along the line to monitor production line performance and to detect potential and actual problems. In a similar fashion, operational BI can monitor critical activities in business transaction workflows and publish results to an operational BI dashboard either continuously, or at regular intervals. For order processing, the dashboard could monitor orders flowing through the system to track factors like order rates, shipment rates, back order and order rejection rates, and to determine order processing elapsed times and order-to-cash efficiency. The results may be used to not only measure and optimize performance, but also to uncover potential problems and errors like incorrect product codes on Web sites, badly designed data entry forms, and so forth. This type of processing can be used to optimize a variety of daily business operations such as call center operations, supply chains management, fraud detection, claim processing, store inventory management, and so forth.
Operational BI can be embedded in business transaction workflows, or may be called as a service in an SOA. There is no inherent reason why this approach cannot also be used in collaborative workflows. The important thing to note about process-centric operational BI is that is analyzes events as they occur in business processes. This event data can be made persistent before it is analyzed, but typically it is analyzed in flight. This improves action times and reduces the amount of data that has to be stored. A summary of the results can be subsequently made persistent in a data warehouse, if required.
The infrastructure vendors are moving toward supporting a process-centric approach to operational BI, as are new BI vendors like Altosoft and Celequest (recently acquired by Cognos). Traditional BI vendors are starting to support an SOA environment, but still have a data-centric mindset.
In summary, we can see that business transaction, collaborative, business intelligence and planning processes are becoming more integrated (see Figure 1). This integration can be achieved through workflows and a business process centric approach that applies not only to business transaction processing, but also collaborative and BI processing. The challenge for both BI vendors and IT staff is adjust to a process centric mindset, and to supply tools and applications that support a business process perspective. From a political and organizational standpoint, process centric approaches will force organizations to dismantle barriers between operational and business intelligence IT groups.