Teradata Magazine Online | Article

Features

Critical Agility

Operational BI generates faster and smarter decisions.

by Colin White

More companies are using operational business intelligence (BI) for faster, smarter decision making and improved efficiency. This trend recognizes the importance of operational BI as a mission-critical capability that adds value.

Working smarter, coupled with increased agility and adaptability, is key to staying competitive in the dynamic, global marketplace. Businesses that leverage the technologies and techniques that enable operational BI will reap substantial benefits.

Strategic and tactical BI approaches ("traditional BI") are well-understood and documented data-centric methods. They employ query, reporting and analysis applications to process operational data that has been consolidated into a data warehouse.

Operational BI is not as well understood, because it is relatively new and can be implemented in several ways. One way is by improving the responsiveness of traditional data warehouse and BI processing. Another is to embed the BI directly in operational processes. These two approaches are not mutually exclusive, however, and are often used together.

Improve agility

Traditionally, BI involves delivering the right information in the right format to the right people at the right time for more informed decision making. In the case of operational BI, however, the focus is on "the right time" and the ability to give users timely information to make faster decisions.

Some people think that right time is the same as real time. While some operational BI applications require such agility, most do not. BI applications such as algorithmic financial trading may need sub-second responsiveness, while others, like inventory management, may require only daily information. The requirements of most operational BI applications fall between these two extremes. Examples include customer call center optimization, order shipment and delivery optimization, just-in-time inventory and Web retail performance.

The agility requirements of these BI solutions vary by organization and user needs. IT implementation costs are also an important consideration. The closer to real time a BI application gets, the more expensive it is to implement because of the additional hardware and software needed to guarantee service levels. Business benefits, therefore, must be balanced against technology costs.

Although agility requirements will determine the types of BI technologies and techniques used, the final application must be flexible enough to support increased agility without major redesign work as business needs change.

SIDEBAR: 3 Types of BI

Business intelligence (BI) used for decision making can be broken into three main types of applications:

Strategic. Such applications help executives as well as business and financial analysts assess progress in achieving long-term, enterprise-wide goals such as increased revenue or market share, reduced costs, better customer retention and improved profitability.
Tactical. These focus on analyzing short-term initiatives within specific line-of-business domains, such as marketing, sales, purchasing or customer service. Helping sales managers optimize their region-wide campaigns is an example of this type of BI application.
Operational. This type features process-centric solutions for monitoring and optimizing specific business processes, such as call center operations, loan processing and inventory management. Operational applications are designed to help organizations manage their intra-day and daily business operations.

BI-driven decision making

Operations change continuously as organizations update business plans and deploy solutions to support processes. To enable this kind of organizational flexibility, traditional BI systems monitor and measure business performance by collecting, cleaning and integrating operational data for evaluation by applications. The applications report on and analyze the integrated data and deliver the results to business users.

These users next collaborate and employ their expertise to decide whether the results are satisfactory and meet goals. If not, they carry out the actions needed to improve matters. The BI system can then be used to measure the effectiveness of those actions. This closed-loop decision-making cycle is illustrated in the figure above. What sets operational BI apart is its ability to increase the agility of closed-loop processing to support daily and intra-day business decisions and actions.

Operational BI can be implemented by improving the speed of traditional BI. Approaches here include:

Updating a data warehouse more frequently and, where appropriate, accessing live operational data to reduce the latency of the information reported on and analyzed
Calculating performance metrics, such as customer lifetime value scores, in advance to reduce analytical processing times
Using automated alerts and rules-driven recommendations to help users make faster decisions

“Organizations often begin operational BI initiatives by looking for ways to enhance the agility of traditional BI or by using existing data for new applications.”

With some projects, traditional BI cannot meet operational decision-making requirements. Data volumes may make it impossible to update the data warehouse and analyze information quickly enough to satisfy agility requirements. For some applications it is not practical, or even necessary, to propagate data from operational systems into a data warehouse. Sensor and hardware alerts, network interactions and messages, stock market trading operations, and RFID tag scanning, for example, could be handled by embedding the BI in the business processes and analyzing the data as it flows through these processes.

SIDEBAR: NCR Corporation

NCR Corporation is a global technology company and leader in automated teller machines, self-checkouts and other self- and assisted-service solutions, serving customers in more than 100 countries. NCR's Worldwide Customer Services organization employs operational business intelligence (BI) as it responds to more than 22,000 service actions/calls per day.

NCR uses a single enterprise data warehouse (EDW) from Teradata as a repository for correlating data across all of its service systems. The EDW, which is updated three times a day, contains information—including the current status—about every service incident. The company's IT services database includes more than 5TB of data representing more than 11 million transactions.

Initially, the EDW was used to do strategic analysis, but more recently it has evolved to support operational analytics, which are used daily within business systems.

The BI application analyzes the history of service incidents and evaluates every situation in terms of how it could best be resolved. Results are then fed back into key operational systems as predictive rules that are used to route and handle calls by NCR's customer care centers.

For instance, if the operational system predicts that an incident can be dealt with remotely, then the call is automatically routed to a second-line customer care professional for resolution. More complicated calls are routed to call management personnel who coordinate the dispatch of a service technician and service parts. Call management personnel use operational systems to determine the parts needed based on the incident code and identify the service technician in the field that has the necessary skills and fastest access to the required service part.

NCR is also extending the system to automatically identify the relative importance of specific parts to resolving the problem.

The operational BI solution enables NCR to optimize its service supply chain and improve business performance by enhancing the customer experience, addressing systemic material issues, lowering inventory investment and identifying improvement opportunities.

The level of success that NCR has achieved is reflected in service statistics such as call wait times of fewer than 28 seconds. Similarly, the company boasts an 85.2% first-visit resolution rate.

The full case study summarized here was published in June 2008 by BeyeNETWORK in the research report "Using Embedded Business Intelligence and Analytics for Near-Real-Time Decisions and Actions," by Judith R. Davis and Colin White (www.beyeresearch.com).

—C.W.

Services

The easiest way to inject BI directly into operations is to break it into a set of services that can be called by business processes and their underlying activities. A BI service can support two types of requirements:

The first is on behalf of a specific business activity. This could include checking a transaction for potential fraud, retrieving an individual's lifetime value metric for call routing in a customer support center, or reformatting a customer address to meet data standards. In some cases, this may be done in conjunction with traditional BI.
The second type of BI service monitors and evaluates the performance of a business process. In this case, event data is passed to the BI service from multiple points in the process workflow. These events are then consolidated and evaluated, and event metrics are produced for delivery to an operational dashboard or other business process. The event analytics can be stored in a data warehouse for subsequent use.

Event analytics determines the efficiency of a complete business process and identifies performance bottlenecks that are affecting operations. Call center performance and order processing typify these types of processes. Stream analytical applications that employ complex event processing technologies to analyze in-flight events with sub-second response times offer a more extreme example. Financial institutions and Web retailing companies use this type of high-volume processing.

SIDEBAR: Embedded BI analytics survey

A recent online survey of BeyeNETWORK users illustrated how organizations were employing embedded business intelligence (BI) analytics:

Forty-seven of the 89 respondents were evaluating, planning or implementing, or had deployed embedded BI analytics applications.
The remaining 42 respondents said their main reasons for not using embedded BI analytics were "Other IT projects have priority" (58%) and "There is not a business need" (31%).
Among those organizations pursuing an embedded BI analytics project, 60% were in the evaluation phase, 13% in planning, 8% in development and 19% in production.
About 62% of those who had implemented embedded BI analytics said it was too soon to gauge their projects' success.
An additional 23% reported their projects were meeting expectations.
The remaining 15% were experiencing difficulties or had projects that were not meeting expectations.

Make the case

Justifying a project involves identifying key operational decisions and business processes that affect performance and benefit from operational BI. Since it focuses on improving specific business processes, this task is simpler because the performance gains are easier to measure, compared with traditional BI projects where the benefits have to be determined for a complete business area or the whole company.

Executives and users can be presented with a case that shows a clear return on investment (ROI) when it can be validated that:

Business users make more informed operational decisions by tightly connecting a specific process to analytical techniques that measure and evaluate the underlying activities involved in running that process.
Increasing the agility of the integrated operational and BI decision-making environment facilitates faster decisions.
A more dynamic business setting enables users to learn, adapt and evolve operational processes based on the analysis of factors that affect performance.

“Operational BI can be implemented by improving the speed of traditional BI.”

How to get started

Organizations often begin operational BI initiatives by looking for ways to enhance the agility of traditional BI or by using existing data for new applications.

New operational BI projects often start with a user-centric approach in which the consumer of the results is a business user of a Web-based dashboard or portal. In this case, the system's agility is usually not real time but several minutes or even hours.

Real-time BI projects, in which the consumer could be a business user or an operational application, require careful planning because of the stringent service levels involved. These types of projects require a sound and scalable hardware and software architecture. They are often developed as a part of an organization's move toward a service-oriented architecture.

Operational BI is one of the fastest-growing areas of BI because it offers the ability to improve performance and efficiency. It can be developed using a wide range of technologies and techniques, many of which can be implemented in conjunction with one another. Which ones are used will depend primarily on the type of BI project involved and the level of agility and responsiveness required.

Colin White is the founder and president of BI Research and is well known for his in-depth knowledge of leading-edge BI and business collaboration technologies.