Data
Synchronization Unifies the Customer View
By
Steve Miranda, Oracle Corp.
What
do John Doe, Johnathan B. Doe and J.B. Doe all have in common? Aside from a
first initial and last name, the answer may be absolutely nothing. They may
be the same person.
Confused? You're not alone. This scenario illustrates the challenge that
many companies face in trying to better understand their customers. In fact,
The Data Warehousing Institute estimated that poor data cost U.S. businesses
$600 billion annually in wasted postage and marketing costs as well as lost
customer credibility.
While many companies have come to re-alize the value of a unified,
enterprisewide customer view, the reality for many is that customer
knowledge is fragmented. Most businesses do not have the knowledge to be
customer-centric and approach projects independently, without leveraging '
or building upon ' their existing customer information. As a recent Gartner
report put it, most companies are in a 'state of denial regarding their data
quality.'
Adding to the challenge is the fact that business environments and processes
are constantly changing and evolving. What is needed is a method that
enables departmental innovation while providing enterprisewide customer
knowledge as a means to delivering a single customer view.
Traditional technology approaches to solving the customer data problem, such
as consolidation of transaction systems, integration projects, business
intelligence systems and analytical applications are only partial (and
temporary) solutions. None address the problem of providing a single
customer view on an enterprisewide level. Point-to-point integration is
costly and unwieldy. As businesses expand through mergers and acquisitions
and as departments upgrade and add systems, the integration challenge
becomes a logistical nightmare, riddled with resource and staff redundancies
and high maintenance costs.
Importing customer data into a data store or warehouse can also be
problematic. Without standardized processes, data from multiple systems are
difficult to reconcile since systems often intersect at multiple entry
points. As the 'John Doe' example illustrates, this can lead to an 'identity
crisis.' Even if this problem gets resolved, the root problems are never
addressed. Data warehouses are unidirectional and can analyze 'historical'
data only, meaning they do not synchronize with the original source to allow
for continued updates.
To be competitive in today's business climate, companies need transactional
systems that operate in real-time. Because so many companies have invested
heavily in 'best-of-breed' solutions to run both front- and back-office
operations, it's unrealistic to expect them to scrap the lot and move to a
single, unified system. The most effective and efficient path for these
companies to follow is to manage customer data centrally, or in a hub. With
the hub serving as a shared resource for all applications and data
collection sites, there is less movement of data, less need for data
preparation and data quality services, and faster data availability.
The hub model provides an optimal solution. Instead of jury-rigging
disparate systems to talk to each other, all customer data are centralized
into a single repository to which the systems connect through a 'two-way
street.' The clean data are then distributed to all systems. This continuous
loop ensures that everyone in the organization is accessing and using the
same accurate customer information. When companies begin relying on
consistent, concise customer data, employees and executives speak the same
language, make better business decisions and provide improved customer
service.
Following is a description of the data hub model and the four steps toward
implementing a hub system: centralize the data, improve the quality,
synchronize with other systems and leverage the 360-degree customer view to
improve all aspects of business.
Centralize
The first step is to consolidate all customer data into a single repository.
This is typically the most difficult phase of the implementation, especially
in larger organizations using multiple home-grown systems. Some companies
have had to reconcile thousands of records for the same customer. Therefore,
it is important for project managers to map out all the systems and fully
understand what information each system needs for daily operation.
The view of a given customer will often vary by department and across
business units. Thus, in building a master customer file, the company must
model each customer as a complex set of relationships and roles (i.e., a
family of companies or multi-person households). For instance, a customer
organization may also be a trading partner, so each role must be noted in
the master file. Similarly, subsidiaries of a holding company should be
rolled into one master customer file to document the umbrella organization's
true purchasing clout and revenue potential (i.e., upsell and cross-sell
opportunities, volume discounts). The hub should model key individuals
within the customer organization. For instance, it's important to know that
a new CEO for a major customer account previously headed sales for another
customer and already has contacts within the company.
The shift to a centralized hub is not an easy task. It often requires
large-scale process changes since all users and systems must align to the
same information model. But once these processes are in the place, the
master customer identities built and the data cleansed and centralized, a
well-designed hub essentially becomes self-sustaining.
Improve
Consolidating data does not automatically create greater visibility. On the
contrary, the information from multiple sources will remain fragmented and
inaccurate unless there's a standardized framework for vetting, cleansing
and organizing the data. This process is known as data quality management (DQM).
Using data quality workbench tools built into a data hub provides companies
with the ability to automate and integrate DQM functions at the point of
collection. Among the features to look for are a configurable duplicate
matching engine (a tool that automatically determines if two different
customer entries are actually for the same person); an extensible merge
routine logic (a set of business rules that govern when and how customer
records are merged when it is determined that there are duplicates) and
out-of-the-box address validation and data certification capabilities
(integration to third-party data sources, such as Dun and Bradstreet, that
help fill in incomplete customer records).
A complete tool set with all of the above characteristics is key to
processing large volumes of data in real-time. For instance, a data
librarian can set up match rules using sophisticated algorithms to flag
spelling errors, formatting irregularities, nicknames and abbreviations.
Data resolution tools offer scoring functions and thresholds to more
granularly define what constitutes a duplicate. Some of today's pre-built
hubs can identify and resolve duplicate data from across more than 100
transactions.
Quality assurance is at the root of file creation, revision and maintenance.
As data enter the hub, duplicates are identified, information is
cross-referenced and verified, and the blended records are either assigned
to the correct master customer file or flagged for human review. This
approach frees individual source systems from all data management, a
traditional source of redundancies and costs.
Synchronize
The third step is to close the information loop. The hub model is designed
as a two-directional information system, so that after data are centrally
processed, the information is pushed back to the systems. This continual
synchronization keeps systems and users across the enterprise notified of
real-time customer events.
To achieve this 'real-time synchronization,' a hub model requires built-in
mapping and control capabilities. While all users access the same customer
data, the various applications may run on different types of information.
Therefore, the hub system should cross-reference and selectively update
systems based on a 'need-to-know' basis. Many hubs are built with an
'editor' to regulate the synchronization process. The hub should support the
programming of both real-time and delayed system updates.
Leverage The 360-Degree View
Once the data have been centralized, improved and synchronized, the hub
begins to deliver on its real promise: a single source of truth across
locations. To achieve this 360-degree visibility, the hub must provide users
with the means to review, search and use the master files.
A universal 'viewer' provides a window across the enterprise, offering an
overall summary of transactional and profile information. Advanced hub
models use analytical tools to model customers in all their complexity. For
instance, customer information can be consolidated from across all global
units, rolled up into hierarchies (i.e., parent households or corporate
umbrellas) or revealed as individual business relationships.
The strategic value of a single customer view is immense. With rising demand
for customized products and services, businesses require more complete
customer knowledge. An organization can implement a 'just-in-time' inventory
approach or lean manufacturing operation with the certainty that it knows
who its customers are, what they are buying and what volume deals are in the
pipeline. Good customer knowledge also drives decisions for product
development, channel strategy and acquisitions. Moreover, accurate customer
knowledge has essentially been mandated by corporate reform laws like
Sarbanes-Oxley, which requires disclosures of material risk to revenue and
profit.
A centralized hub model also represents a smart investment in the future. As
a closed information loop, the hub keeps getting 'smarter.' As data accrue
on new and existing customers, reporting accuracy improves and analytics
become more valuable. Faster response times allow companies to operate more
efficiently while improving the customer experience.
The need for better customer data is no longer a secret. Data-quality
software is now a $600 million a year industry, according to Forrester
Research. As more companies recognize the need for good data, customers are
starting to notice those that don't. 'John Doe' can only tolerate so many
flyers in his mailbox from the same company before he says, 'Enough.'
A centralized data hub is an ideal way to solve the customer data problem by
enabling companies to gain a single source of real-time 'customer truth,' no
matter how diverse their IT environments are.
Steve Miranda is vice president of Applications Development at Oracle
Corporation (www.oracle.com).
If you are interested in purchasing reprints of this article (in either
print or HTML format), please visit Reprint Management Services online at
www.reprintbuyer.com or contact a representative via e-mail at [email protected]
or by phone at 800-290-5460.
[
Return To The
September 2004 Table Of Contents ]
|