That business intelligence (BI) is one of the leading technology investment areas
for organizations today comes as no surprise. Following major investments
in enterprise applications like enterprise resource planning (ERP) and customer relationship management (CRM),
organizations seek to better leverage data captured by these systems.
The renewed interest in business intelligence comes at a time when the very concept is evolving. Traditionally,
BI has been seen as "decision support." Business analysts use technologies such as query, reporting, and analysis to develop information that can be delivered to managers (i.e., those conventionally empowered to make decisions). Business intelligence systems have long been
used to report and measure business performance "after the fact" — i.e.,
after the events in a business process have occurred. Decision-making processes, oriented toward prioritizing future actions, are outside the scope
of a traditional BI system.
However, BI as we know it is expanding simultaneously in two directions:
to the rank and file
Thanks to dashboard and scorecard technology, the trend today is to deploy BI to all knowledge workers, not just upper management, especially as a tool to measure current performance levels against corporate goals or targets. This is in line with the increasing decentralization of decision-making authority in many organizations.
BI is undergoing another transformation, perhaps more slowly, but no
less inevitable or inexorable, into business processes. The trend will be to incorporate BI directly into operational applications, providing a level of consistency in the execution of repeatable, operational decisions across a broad group of decision makers. Embedding BI in a business process represents the logical
evolution of business intelligence — from after-the-fact business intelligence reporting, to analytics, to
in-line, context — sensitive analytic applications that guide the user to make effective decisions.
Why Embed BI in a Business Process?
Consider the following example of such a repeatable, operational decision: A manufacturer receives a large order
for goods from a retailer, but the order exceeds the credit limit for that customer. Should the manufacturer turn down the order or raise the credit limit? If the organization adheres rigidly to the credit limit, they could be missing a significant opportunity. On the other hand, it would be reckless to routinely raise credit limits when the chance of receiving payment is small.
Think about this example from the perspective of an order management application. The application automates the business process by defining the series of steps that must be followed
to execute the transaction. Before committing the transaction, the system checks the customer's credit limit. In
a traditional scenario:
- The application could simply abort the transaction and send a message that the order cannot be processed due to a credit violation, or
- An exception could be declared, and the issue would then be sent to a manager who must make the operational decision.
In each case the automation ceases, and a manual decision-making process takes over. Decisions such as this one will be presented to different managers over the course of a business period. But what assurance is there that each occurrence of this operational decision will be optimal? Will each manager follow the same process in reaching his or her decision?
The New BI Frontier: Automating Decision Processes
Companies are beginning to realize that without a defined process and a level of automation, they are assuming
a level of risk. If the process is not consistent, might the company risk being sued for bias by an angry customer, or prosecuted by a government authority for not complying with a credit regulation?
For reasons of achieving maximum performance and also for compliance, an agile company will be looking to incorporate intelligence in such operational processes. How do you bring
information to the core of the decision and actually look at and evaluate decision alternatives so that you can make an optimal choice? From the perspective
of the applications industry, decision-oriented processes are the next frontier for automation.
The Two Roles of BI: Information Delivery and Decision Optimization
The challenge of optimizing decisions does not eliminate the need for delivering information. Intelligent
decision making will always require information. But that information must be timely (available in the
context of a knowledge worker's activities) and relevant to the situation at hand. The upper half of Figure 1 illustrates the information delivery role of business intelligence.
|BI's Dual Roles: Information Delivery and Decision Optimization
Traditional BI places a premium on delivering information with ever-greater speed. In order to ensure accuracy, this pursuit of timely information must be combined with a careful focus on data quality. Automating the gathering,
integrating, and reporting of data can help improve the currency of the information that is delivered. This helps a finance department to close the books (i.e., deliver financial reports) faster and provides a sales manager an analysis of sales by region in less time.
But traditional BI is not engaged in the business
processes upon which decisions are made. Getting
more accurate information faster improves business
agility, but only when this feedback can be applied
to make ongoing corrections to business operations
or to make intelligent, in-process decisions about
alternatives. The steps in the workflow for such
operational decisions are shown in the lower half
of Figure 1. What adjustments might be possible with such a system? For example:
- A retailer can change the price for selling seasonal goods by analyzing recent shifts in demand
- A marketer can make a new offer available to a targeted customer segment by analyzing recent
- A manufacturer can raise a customer's credit limit
Before such an operational decision is made, predictive models evaluate alternative courses of actions (hypotheses) to project the probability of an outcome. This intelligence is applied (via an operational decision) when a merchandiser marks down a price or a marketer tailors an offer to a particular customer or customer segment.
Taken together, the upper and lower halves of Figure 1 represent a closed-loop process that applies feedback to adjust business operations. But today, using traditional BI approaches, not all of the steps are automated or linked. There is a disconnect between the
workflows in the upper and lower halves of the diagram. The separation of BI from the decision-making process is the learning gap.
Bridging the Gap
How can this business process gap be addressed? Business rules must be
developed to handle exceptions in a uniform way. Rather than kicking the user out of the application (requiring manual intervention), the analysis of relevant information via business
intelligence should drive the direction of the process and guide the user as decisions are presented.
In the case of our credit decision example, the order management process could be extended to include the analysis of a customer's credit history that would yield a credit score. A predictive model could be built to take these scores and indicate the likelihood of the customer making payment if the credit limit were raised.
Armed with this information, a credit manager would no longer have to rely on a gut feel, but on a structured process that seamlessly combines transactional and analytic steps and is defined and certified by the organization to handle this type of a situation.
The organization is able to capture and promulgate "best practices" for decision making, thereby making itself more responsive to customers, more flexible, more agile, and more competitive. Periodically, the effectiveness
of the decisions can be monitored, and the rules can be adapted or processes evolved. The result is a dynamic order management process with more functionality and a higher degree of automation — thereby more efficient and more compliant.
Supporting Decision Processes via Composite Applications
Automating these processes is admittedly challenging, as they will require a mix of technologies (event monitoring, modeling of decision alternatives, and business rules) along with domain expertise for each decision type.
Composite applications, such as SAP xApps, have emerged as one way to support such decision-making processes by embedding business intelligence. A composite application — an application with a business process manager layer that sits atop a group of related applications and coordinates the workflow between them — orchestrates a business process flow, invoking multiple code sets and types of applications inside and outside the enterprise. The workflow can be adapted using a business-modeling tool that is accessible by a business rather than an IT professional.
The business process flow in a composite application connects steps that are implemented via multiple underlying applications or services:
- Transactional: Managing an interaction, be it with a customer, supplier,
or employee — for example, communicating with a consumer seeking to order a product over the Web
using a shopping cart or e-commerce application. The result is a record of a business event such as a
sale, which is committed to the operational database.
- Analytic: Monitoring and analyzing information — for
example, sending alerts of a significant event involving
a high - value customer, triggering a proactive communication
or specific response to an ongoing interaction.
- Collaborative: Making decisions on rules to be used when specific events, conditions, or
states occur — for example, deciding on offers to be made for customer retention, cross-selling, and promotions based on analytics such as a customer's
profitability or likelihood to churn.
Composite applications extend the reach of business intelligence and of traditional enterprise applications, providing consistency and optimal
business value via the automation of repeatable, operational decisions.
The Future of BI
Why should anyone make an investment in BI that provides more guidance around decision making? The two greatest
motivators today are improved business efficiency and regulatory compliance. As business intelligence bridges that "learning gap," companies will realize a reduction in costs and an increase in productivity, plus they will satisfy their auditors and regulators.
Companies first must look at their business processes where BI could be applied to make a difference. For maximum return, look for processes that:
- Are repeatable; one-offs are not worth the effort
- Produce a measurable benefit, either an improvement in the process or an improvement to your business
- Must fulfill compliance requirements — if there is regulatory hook, so much the better
- Are industry-specific for more bite-size operational decisions with immediate impact
Begin looking now for the right match of processes and BI technology that can not only supply the right data, but also infuse it into each step of the decision-making process.
Expect that BI will increasingly allow you to close the current gap between the collection of
information and the ability to act on that information. The future of BI will be in allowing your
company to predict and evaluate decision alternatives in a consistent manner — and ultimately make
Henry Morris is
Group Vice President and General Manager
for Integration, Development, and Application
Strategies at IDC, a worldwide research
firm with headquarters in Framingham,
MA. He started the Analytics program
at IDC and coined the term "analytic applications" in
1997. He has written extensively on analytic
applications, decision-centric business
intelligence, and the ROI of analytics
projects. Dr. Morris earned his B.A.
with distinction from the University
of Michigan and his Ph.D. in philosophy
from the University of Pennsylvania.