Moving Forward with the On Demand Real-time Enterprise

Chapter 3. Architectural considerations 99

data warehouse analyses. A similar approach can be used to synchronize

customer analyses that are sourced from both systems.

Figure 3-8 MDM and analytics

3.3.3 Analytic applications

Analytic applications are not a new thing. They have been around in one form or

another for a number of years. However, now that we are evolving further into the

realm of data warehousing and business intelligence the time has come to dust

them off, enhance them, automate them, and use them as another critical

capability in the world of analytics, or business intelligence as we prefer.

For many years we have asked people to dive into the depths of the data in a

data warehouse, analyze it, and use the results to resolve all the issues of the

enterprise. Unfortunately that has not happened, at least to the degree which we

would have liked. And the likelihood is that it will not happen, because there is

just too much data and too few people with the domain knowledge required to

really understand and analyze the data. So, we need help.

Analytic applications to the rescue. What we an do is glean some of that domain

knowledge from the gifted few, and instantiate it in the analytic application. And

what does not get put into the application, should find its way into a rules base.

Alert Message:

Product

PROD01 return

costs too high

Call Center Data

ETL Processes

Master Data Management

GetNormalizedProduct()

Master Product Information

Original Normalized

prod1a prod01

prod1b prod01

Data

Normalization

Operational Systems

CRMERP

Products

PROD01B, XXL,

500, …

CRM Data

PROD01A, XXL,

500, …

Business

Process

Management

Business Monitor

Action Manager

BPC

Common Event Infrastructure

Alerts

Monitor Data

… PROD01B, …XXL

PROD01:

Historical Returns Analysis

Operational & Analytical Dashboard

Inconsistent

Product Metadata

Transformations

PROD1A

Æ PROD01

GetNormalizedProduct()

Enterprise Data Warehouse

Sales

PROD01, 1000,

…

100 Moving Forward with the On Demand Real-time Enterprise

What we are doing is leveraging the domain knowledge and, where applicable,

enabling it to be automated. Here is a way to enhanced efficiency and

effectiveness.

With this domain knowledge and rules base, an analytic application cannot only

access data, it can analyze it and take action - even if that action is as simple as

sending an alert to a human. Or, we can further extend that application capability

to enable the application to better equip the human. That is, with embedded

knowledge it can help guide the human. Rather than simply alerting the human,

the application can offer plans of action. For example, it could recommend

looking at a particular area of the data warehouse, or other related data.

Basically what you are doing is embedding the knowledge and problem solving

capability of a number of skilled analyst into the analytic application. Although not

perfect, it is a giant step in the right direction. It is enhanced BI in a real-time

environment. Here is another characteristic of a real-time enterprise?

In-line Analytics

Today, business intelligence applications are used in a relatively small

percentage of corporate environments. This is particularly true when considering

the front-line workers who are executing the business processes. One of the

primary reasons is that business intelligence applications may be more difficult to

use because they need data at the right time and in the right context to enable

them to make decisions, and many times there are also complex analytics

involved. Therefore it is critical that businesses embed business intelligence

tightly into the business processes - in the form of in-line analytics in their

applications.

In-line analytics represent the ability to provide analytical services at the point of

decision, to optimize decision making. This means that analytics need to be an

integral part of the business process. As a user or system is at a point of decision

within the business process, analytics are delivered in the appropriate business

context.

Types of in-line analytics

In this section we discuss three ways for the delivery of in-line analytics:

1. System-driven analytics refer to a system that programmatically consumes

an analytical service for decision-making. That is, decisions are made using

information as a service. As an example, consider a customer who wants to

place an order. During the process, the system will programmatically suggest

payment methods, using a data mining solution, and based on a score from

the customer payment history.

2. User-based real-time analytics are analytics provided within a process step

that requires user intervention. It is critical that the decision is made in