Columns
Understanding analytic applications
- By Wayne W. Eckerson
- October 31, 2002
The term ''analytic application'' has emerged as an important new business
intelligence trend. Yet many business intelligence professionals wonder if this
term truly signifies something new, or whether it simply repackages existing
analytic technologies and processes under a new banner.
The truth is that the market for analytic applications is both new and
old.
What's old?
For the past 10 years, business intelligence
professionals have employed analytic technologies and products -- including data
warehouses, query and reporting tools, OLAP products, data mining tools and
algorithms, and visualization techniques -- to deliver information and insight
to business users. In most cases, organizations use these tools to create access
methods and reports for specific functional areas, such as sales, marketing,
finance and manufacturing.
Many business intelligence professionals may consider the results of these
projects to be analytic applications. And they are right -- to a degree. These
handcrafted solutions apply analytic technologies to business needs in specific
functional areas.
So, what's new about an ''analytic application'' in the year 2002?
What's new?
Four factors differentiate today's analytic
applications from previous generations of analytic tools and applications. TDWI
calls these the four ''P's'' of analytic applications:
1. Perspective. A true
analytic application contains some level of domain knowledge about a functional
area, such as sales, marketing or manufacturing, in a particular industry. The
best analytic applications embody industry best practices represented as key
performance indicators or metrics within a set of pre-defined reports or report
templates.
For example, a procurement analytic application might provide a set of
reports that help purchasing managers optimize spending on materials for a
manufacturing plant. The spending optimization reports will highlight
best-practice metrics, such as aggregate spending per supplier on a global
basis, average order value and percentage of order volume purchased outside
existing contracts.
''The real benefit of an analytic application is the domain knowledge or
intellectual capital it contains about a functional area,'' said Bill Schmarzo,
vice president of analytic applications at DecisionWorks Consulting and a TDWI
faculty member who instructs on analytic applications at TDWI conferences.
2. Process. Another trademark feature of an analytic application is that it
is ... well ... an application. Just as operational applications walk users
through a pre-defined business process step by step, analytic applications do
the same for decision-making processes.
An analytic application is not just a bunch of reports, or an analytic tool
to perform ad hoc queries or create reports. It is a real application that
provides transparent support for the analytic processes that individuals and
groups use to analyze data, make decisions and act on plans. These analytic
processes can be procedural, contextual, collaborative, event-driven,
transactional or evaluative.
In the past, organizations gave users an analytic tool, provided some
training and hoped for the best. The result was usually failure or lots of
underutilized software. A report by Nigel Pendse and Survey.com found that
organizations never deploy an astonishing 39% of the OLAP licenses they
purchase. (See http://www.survey.com for
''The OLAP Survey,'' by Nigel Pendse, July 2001.)
Analytic applications promise to reverse this trend by embedding analytics
into the fabric of users' daily business processes and tasks. Users don't have
to wrestle with a tool to access, analyze and act on information. They simply
use an analytic application that supports their decision-making style, business
processes and collaborative activities in a seamless, intuitive fashion.
3. Packages. In the past, organizations had to stitch together multiple
products and components using hand-written code to create an analytic
application. Today, many vendors offer packaged analytic applications that
pre-integrate analytic and data warehousing components, including data models,
ETL tools, meta data, analytic tools, reports and portals.
These packaged applications typically provide 65% to 85% of a complete
solution, greatly accelerating and simplifying the deployment process. Most
analytic packages are tailored to specific functional areas, such as sales or
marketing, or to applications in vertical industries, such as retail
merchandising and assortment analysis. These packages embed domain knowledge of
specific functional or vertical applications in data models and reports.
4. Platforms. Contrary to public opinion, analytic applications are not just
packaged solutions. Today, organizations can just as easily build an analytic
application as buy one. That's because an emerging class of vendors offers
specialized tools to rapidly build custom analytic applications on top of an
existing data warehouse or data mart.
These tools -- which TDWI calls analytic development platforms -- enable
developers or savvy business users to build custom analytic applications using
pre-defined components, services and starter kits in a graphical environment
that minimizes coding and facilitates rapid prototyping and deployment. (There
are also data warehouse development platforms, sometimes called packaged data
warehouses, that automate the design, deployment and maintenance of a data
warehouse.)
In contrast, most analytic tools (a.k.a. decision support tools or business
intelligence tools) are hard-wired monoliths with a vendor-supplied look and
feel -- what you see is what you get. Administrators cannot easily modify the
vendor's GUI, add new functions, modify or extend existing functions, or
dynamically personalize the end-user environment (GUI, functionality or views,
for example,) to fit the user or group.
The complete solution
In summary, an analytic application is a
domain-specific analytic solution that integrates a diverse set of data
warehousing and analytic tools that organizations previously had to
painstakingly stitch together.
New packages and development tools enable organizations to rapidly deploy
analytic solutions that address the unique information needs of knowledge
workers in specific departments or lines of business. As apps, these new
products provide built-in support for decision-making processes that knowledge
workers use to access, analyze, collaborate and act on information.
Definitions
Analytic
application: An analytic application enables business users to
access, analyze and act on information in the context of the business processes
and tasks they manage. It embeds domain knowledge that supports the unique
information requirements of users in a specific department or functional area.
An analytic application is a complete solution that usually leverages a data
warehousing environment, embeds analytic tools and employs business process
logic. You can either build or buy an analytic application.
What it's
not: An analytic tool.
Packaged analytic application: The ''buy'' option --
a vendor-supplied package that provides domain-specific analytics. It contains
an integrated set of analytic tools, data models, ETL mappings, business
metrics, pre-defined reports, and ''best practice'' processes that accelerate the
deployment of an analytic application in a given domain or across multiple
domains.
What it's not: An analytic tool.
Custom analytic application: The ''build'' option -- an analytic application
that is primarily built using tools, code or customizable templates to provide
the exact look, feel and functionality desired by an organization for its
analytic environment.
What it's not: An analytic tool.
Analytic Development Platform
(ADP): A development environment that enables
developers or savvy business users to build custom analytic applications using
predefined components, services or starter kits in a graphical environment that
minimizes coding and facilitates rapid prototyping and deployment.
What it's not:
An analytic tool.
Business analytic tool (or analytic
tool): A tool that provides query,
reporting, OLAP or data mining functionality for end users, but offers little or
no ability to extend or substantially customize functionality or the end-user
environment.
What it's often called: A business intelligence tool or decision
support tool.
Business intelligence: The processes, technologies and tools needed to turn
data into information, information into knowledge, and knowledge into plans that
drive profitable business action. Business intelligence encompasses data
warehousing and business analytic tools, as well as content and knowledge
management.
What it's not: An analytic tool.
About the Author
Wayne W. Eckerson is director of education and research for The Data Warehousing Institute, where he oversees TDWI's educational curriculum, member publications, and various research and consulting services. He has published and spoken extensively on data warehousing and business intelligence subjects since 1994.