INSIGHTS » Decision Management is about using predictive analytics, business rules and optimized strategies to make smarter decisions and eliminate the gaps between insight and action. The FICO Decision Management architecture provides a common service-oriented foundation for rapidly bringing analytics-driven decisioning into more areas of the customer lifecycle—and for connecting decisions across it in ways that create more value for financial institutions and their customers. The new architecture enables IT organizations to help business units precisely balance risk and opportunity in the millions of day-to-day customer decisions and transactions that determine overall performance. It provides IT with faster project implementations, easier integration of third-party products and more agile ways to help business units respond to competitors and market change. This white paper presents an overview of the FICO Decision Management Architecture and its benefits to financial services institutions, followed by a detailed tour of the major components. It also discusses the current migration of FICO’s industry-leading decisioning applications and tools onto the architecture. As part of a new Decision Management application suite, these interoperable solutions will bring clients richer, more powerful originations, customer management, collections & recovery, and fraud management capabilities than ever before, combined with increased configuration flexibility, simpler maintenance and lower total cost of ownership. Number 15—March 2009 www.fico.com Make every decision count TM The architecture described in this paper is the foundation for new releases of FICO Decision Management applications, such as FICO™ Falcon® Fraud Manager and FICO® Debt Manager™ solution. Product roadmaps and similar marketing materials should be considered forward looking and subject to future change at FICO’s discretion. Future functionality, features or enhancements as shown are FICO’s current projections of the product direction but are not specific commitments or obligations. An Architecture for Smarter Decisions A common architecture for all FICO customer Decision Management applications helps financial institutions improve risk management and performance across the customer lifecycle
19
Embed
An Architecture for smarter Decisions · Decision Management is about using predictive analytics, business rules and optimized strategies to make smarter decisions and eliminate the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
insights »
Decision Management is about using predictive analytics, business rules and optimized strategies to make smarter decisions and eliminate the gaps between insight and action.
The FICO Decision Management architecture provides a common service-oriented foundation for rapidly bringing analytics-driven decisioning into more areas of the customer lifecycle—and for connecting decisions across it in ways that create more value for financial institutions and their customers. The new architecture enables IT organizations to help business units precisely balance risk and opportunity in the millions of day-to-day customer decisions and transactions that determine overall performance. It provides IT
with faster project implementations, easier integration of third-party products and more agile ways to help business units respond to competitors and market change.
This white paper presents an overview of the FICO Decision Management Architecture and its benefits to financial services institutions, followed by a detailed tour of the major components. It also discusses the current migration of FICO’s industry-leading decisioning applications and tools onto the architecture. As part of a
new Decision Management application suite, these interoperable solutions will bring clients richer, more powerful originations, customer management, collections & recovery, and fraud management capabilities than ever before, combined with increased configuration flexibility, simpler maintenance and lower total cost of ownership.
Number 15—March 2009
www.fico.com Make every decision countTM
The architecture described in this paper is the foundation for new releases of FICO Decision Management applications, such as FICO™ Falcon® Fraud Manager and FICO® Debt Manager™ solution.
Product roadmaps and similar marketing materials should be considered forward looking and subject to future change at FICO’s discretion. Future functionality, features or enhancements as shown are FICO’s current projections of the product direction but are not specific commitments or obligations.
An Architecture for smarter Decisions A common architecture for all FICO customer Decision Management applications helps financial institutions improve risk management and performance across the customer lifecycle
www.fico.com page 2
An Architecture for Smarter Decisions
insights » insights »
The Need for Software » Architectures to Evolve
The intensity of competition in financial services markets in recent years has put IT organizations
in the spotlight and under pressure. They are seen as the enablers, providing institutions with
the means to outdistance rivals by making smarter decisions, spending less and acting faster in
the face of both risk and opportunity. At the same time, IT organizations are themselves under
pressure to do more with less—to implement technologies that will supply more advantage,
quicker, despite constrained budgets.
The current global financial crisis multiplies these conditions and stakes many times over. While
some institutions are responding to adversity with retrenchment, others know that making
smarter decisions is more important now than ever before. Producing more with less is at the
very core of Decision Management.
Analytics-driven decisioning pinpoints where to focus constrained resources for the biggest
performance and competitive gains. It is the way to create efficiencies and cost savings that both
ease today’s budget pressures and help propel tomorrow’s growth.
For financial services IT organizations to continue moving ahead to meet rising expectations and
challenges, however, software architectures must evolve to facilitate the effort. The monolithic
systems that have so far brought improvements to particular areas of bank operations can’t
provide the agility needed in today’s markets nor the high return on investment from IT now
required. Banks can no longer afford long implementation time frames or constant IT
involvement in executing business responses to market change. IT organizations need the
means to empower business users to make many of these changes themselves, and efficiently
govern that process.
Another imperative for leveraging IT efforts is to reduce the current drain on resources from
maintaining disparate platforms across various areas of customer lifecycle decisioning—
expenditures that deliver limited net business return. IT organizations should be freed from the
necessity of building and maintaining the costly point-to-point interfaces currently required to
overcome lack of interoperability among in-house systems and between them and third-party
systems.
New software architectures based on common, standards-based components are needed to
allow IT to shift its focus from connecting systems to connecting decisions. By providing the
means to connect decisions from originations to collections as well as across business units,
IT will enable banks to decrease the overall risk and increase the overall value of customer
relationships—an expenditure that promises a very large net business return indeed.
“Look for opportunities to turn IT efficiency pressures into ways to improve organizational processes and procedures…Determine elements of competitive advantage and ensure that they are strengthened, not reduced.”
—Gartner 2008, IT Efficiency in Financial Services and Insurance
“The pressure to improve risk management is rising, since most FSIs have yet to implement an integrated approach to risk management across customers, products, and locations.”
—The TowerGroup Top 10 Financial Services, Sept 2008
www.fico.com page 3
An Architecture for Smarter Decisions
insights » insights »
FICO is helping financial services IT expand its important role as a competitive enabler in today’s
tough business environment. We are bringing our advanced technologies and industry-leading
applications onto a new Decision Management architecture that allows IT to accelerate the
implementation of data-driven customer decisioning applications and generate more value
from them by connecting decisions across the customer lifecycle. Based on service-oriented
architecture (SOA) principles and adhering to industry standards, the new architecture also helps
IT simplify maintenance, since FICO decisioning applications from originations to collections are
incorporating the same core technologies and sharing common components.
The FICO Decision Management architecture also increases agility by enabling IT—and
designated business users—to modify decision logic quicker and more efficiently than ever
before, without affecting other parts of the applications. In fact, the new architecture advances
FICO’s efforts over the past decade to decouple decision logic from application code. We’ve
increasingly moved FICO’s applications toward increased separation and addressability of
decision logic, while providing the world’s leading business rules management system (BRMS)
to enable financial institutions to build their own decisioning applications.
The FICO » Decision Management Architecture
FICO Decision Management applications are built on the same core technologies, simplifying maintenance tasks and training requirements
The service-oriented approach facilitates integration of third-party products with decisioning applications
State-of-the-art decisioning capabilities can be layered onto existing systems, extending their lifespan and ROI
Policies and other business rules can be shared across Decision Management suite applications; rule updates made in one place are simultaneously propagated across all applicable decision areas
Single logical data model for all Decision Management applications and unified operational, analytic and reporting databases speed implementations, support data sharing and interoperability, and enable records to be updated once from a central location
The Decision Management architecture gives IT the flexibility to swap its own Data Services for those supplied by FICOapplications, and thereby locate data stores at any client-preferred location
A single, standardized interface provides access to more than a thousand financial services data and service providers
Traditional financial services software FICO Decision Management suite applications
Applications in different decision areas are built on disparate core technologies, multiplying maintenance tasks and training requirements
It’s often difficult and time-consuming to integrate third-party products with existing applications
IT has to engage in costly “rip and replace” when legacy systems are unable to meet new decisioning requirements
Policies and other business rules are often coded and maintainedin multiple places, resulting in redundant work and opportunitiesfor inconsistency and error Inconsistent data definitions and separate databases in use across the customer lifecycle slow application implementations, create barriers to data sharing and cause redundant work
Applications often dictate where data has to be managed, preventing IT from locating data stores based on organizational requirements and efficiencies
Point-to-point interfaces must be built and maintained for access to each third-party data or service provider
Financial services IT needs a new approach:
www.fico.com page 4
An Architecture for Smarter Decisions
insights » insights »
Now, as FICO applications migrate onto the new architecture and become part of the new
Decision Management suite, they gain the full power and flexibility of this best-in-class BRMS—
completely integrated into interoperable solutions for originations, customer management,
collections and recovery, and fraud management.
Decision strategies in one area may check status and activity in other areas, such as an originations application checking to see if the applicant for a new line of credit has a current delinquency
Predictive models and other analytics can be imported directly into rules-driven decisioning processes for one or more Decision Management applications
Decision logic is decoupled from other application code and managed as business rules, so IT can make quick policy and process changes or empower business users to do it (using controlled Rules Management Applications)
Best-in-class “champion-challenger” facilities are available across all Decision Management applications for systematic testing and continual improvement of decision strategies
Consistent tools and frameworks can be used to facilitate the optimization of decisions across the customer lifecycle
All Decision Management applications share a consistent, automated means of capturing decision outcomes and systematically tracking them against predictions and strategies for performance management
Adaptive technologies can be layered onto existing models to increase their sensitivity to changes in the current production environment and maintain high predictive performance over longer periods
Traditional financial services software FICO Decision Management suite applications
Customer decisions sometimes don’t consider the whole picture, such as when a credit line increase is offered on one account while another account for the same customer is delinquent
Predictive models aren’t being used for decisioning as much as business units would like since deployment usually requires models to be recoded for each application’s operational environment
Responsiveness to competitive, market and regulatory change may lag because decision logic is intertwined with other application code and modifying it therefore requires reprogramming
Improving decision strategies is often a trial-and-error process because systematic methods for simulating, testing and comparing strategies are not available in all areas of the customer lifecycle Optimization could provide large competitive and performance gains, but its impact is limited because of the time and expense involved in applying the technology to decisioning
It’s difficult to evaluate the performance of predictive models and decision strategies since the context in which they were made is usually not recorded and tracking outcomes may requiretime-consuming, unreliable semi-manual methods
In dynamic markets, periodic retraining of predictive models may not be adequate to keep up with rapid changes in the production environment
IT clients need a new approach:
www.fico.com page 5
An Architecture for Smarter Decisions
insights »
The FICO Decision Management architecture brings financial institutions the key advantages
typically associated with SOAs:
Quicker response to change• . In general, SOAs accelerate organizational responsiveness
since the individual services making up the software can be modified independently of
each other. In the case of the FICO Decision Management application suite, the
power of this approach is manifest in the ability to quickly modify the logic driving the
Decision Services at the heart of our applications. Because decision logic is managed
separately from other application code and exposed for easy editing, financial institutions
can change their decisioning policies and processes as fast and frequently as necessary.
Where it’s happening (key participating services):
What’s happening:
Agent or online applicant requests credit line
Originations Case Management
2 Collections Case Management
3
Fraud ScoringDECISION SERVICE 4
5Credit ScoringDECISION SERVICE
Originations ScoringDECISION SERVICE
6
Originations ApprovalDECISION SERVICE 7
Originations Case Management
8
Fraud Case Management
4a
Application Data ValidatorBUSINESS SERVICE
1
Credit Application COMPOSITE BUSINESS SERVICE
Create case
Check for delinquency
Score
Score
Score
Validate data
Publish
Decide/Assign credit limit
Submit application, Deliver decision
Data ValidatorDECISION SERVICE
Credit Application, a composite business service, receives request for new credit line and orchestrates response:
1. Calls the Application Data Validator business service, which submits the application data to the Data Validator decision service. It receives back a successful validation or a failure with reason codes. The Credit Application composite service uses these codes and business rules to generate a request for missing or corrected data back to the application orginator.
2. Calls Originations Case Management to create a new application case.
3. Checks with Collections Case Management (which is Business Process Management driven) to see if the applicant has a currently delinquent account.
4. Calls the Fraud Scoring decision service to see if the applicant has any accounts with recent fraudulent activity. Application Fraud Models and identity theft analytics may also be invoked.
4a. Scores above the client-defined threshold generate a fraud case, which is directed by business rules to the queue of a fraud analyst.
5. Calls the Credit Scoring decision service, which pulls credit bureau scores via the Data Access Layer’s data exchange services.
6. Calls the Originations Scoring decision service, providing it with application data and component scores. Receives back an approval score.
7. Calls the Originations Approval decision service, providing it with the approval score. Based on the score and business rules, the service generates a decision (approve, decline, request more information, approve with conditions, etc.) and assigns an initial credit limit.
8. Publishes the decision event and context, which is received by all interested parties, including Originations Case Management, Performance Management and unified operational, reporting and analytic databases. Also initiates actions, such as generation of approval message to the credit application originator.
Portal
CallingApplication
Identity Management
System Management
. . .
DESIGN TIME EXECUTION TIME CLIENT ENVIRONMENT
Common Browser-Based GUI
Strategy Design
Services Layer
Common Data Model
ModelManagement
Analytics OLAP/Reporting Operational Data
RuleManagement
Simulation &Optimization
Configuration & Deployment
Rules &
Analytics
Client Data Store
3rd Party Data
Data Access Layer
Software ServicesDecision Services
OrchestrationBusiness Process
Management
Case Management
Business Services
Security
Enterprise ServicesPerformanceManagement
Learning
Rules ModelsAdaptiveEngines
Service Invocation
Event Management & Monitoring
Provider Access
Enterprise Service Bus
www.fico.com page 10
An Architecture for Smarter Decisions
insights »
Strategy Design
The Strategy Design Layer provides design time tools for modifying the decision logic that
drives Decision Services as well as for configuring Business Services, such as workflows and case
management queues. There are two levels of design time tools:
A complete integrated development environment (IDE) is provided for IT and other technical 1.
users. The IDE is based on the Eclipse open source development platform and the FICO™
Blaze Advisor® system, the world’s leading business rules management system. It provides
extensive control and flexibility for determining how Decision Services operate, and for adding
predictive analytics and optimization to them. It also provides the means to create new
Decision Services.
Prebuilt browser-based Rules Management Applications (RMA) are shipped with each application 2.
to enable nontechnical users to quickly modify business rules and workflows without need for
IT assistance. These RMAs are easily modified and updated, and new RMAs can be created
from the IDE.
Underlying both levels is a complete set of management controls for enforcing the orderly
processes that must balance and support flexible strategy design across large enterprises. These
include an XML-based rule and model repository that performs tracking, version control and release
management as well as tools for testing rules and simulating their impact on decisions.
Model management:The architecture provides a single environment spanning the customer lifecycle for implement- •
ing predictive models and other types of analytics in decisioning processes. It also incorporates
consistent, systematic means for evaluating the accuracy of their insights (see Performance
Management below).
The IDE enables analytic models to be imported directly into rules-driven decisioning processes, •
which can be deployed as Decision Services into virtually any operational environment.
Models imported in PMML (Predictive Model Markup Language) format can be viewed and •
edited in the IDE. This includes custom models produced with FICO™ Model Builder. This
tool, used both by our own analytic teams and by clients developing their own analytics,
provides a comprehensive set of modeling capabilities (linear regression, logistic regression,
The pluggable security architecture also enables clients to add their own security features, such •
as third-party authentication software and credential mapping facilities.
www.fico.com page 16
An Architecture for Smarter Decisions
insights »
In addition, the architecture supports the full range of WS-* security standards, including:•
. WS-Security 1.1 to encapsulate XML within a SOAP document
. WS-Security Policy 1.2 to define a standard way of securing messages exchanged between
web services and clients
. WS-Secure Conversation 1.3 to implement the policy from WS-Secure Policy
. WS-Trust 1.3 to provide security tokens based on the information from WS-Secure
Conversation
. WS-Federation 1.1 to implement the authentication from WS-Trust
Enterprise services:FICO is using standard interfaces for integrating with enterprise services provided by third- •
party products clients already have installed. These include LDAP (Lightweight Directory
Access Protocol) for directory services, JMX (Java Management Extensions) and JSR 77 for system
management services, and UDDI (Universal Description Discovery and Integration) for web
service discovery services.
For clients that do not have products currently installed, there is an option to use the IBM Tivoli •
and WebSphere products to provide these capabilities.
Data Access Layer
The Data Access Layer consists of the Data Services that manage access to the unified •
operational, analytic and reporting databases for Decision Management suite applications.
The layer also provides access to data stored externally, both in client databases and third-
party sources.
Each Data Service manages a set of logically grouped database tables and/or XML data schemas. •
A Customer Data Service, for example, stores and retrieves information such as customer ID,
name and address. A Payment Data Service stores and retrieves payment and payment history
information. In addition to performing basic CRUD (create-retrieve-update-delete) functions, it
carries out specialized operations to optimize access for a specific business purpose, such
as retrieving only the necessary data for a particular type of collections contact.
Data Services are accessible to all Decision Management suite applications via a standard defined •
interface (“contract” consisting of data definitions and invocation mechanisms).
Clients can configure the Data Access Layer to designate which attributes within the Common •
Data Model are to be encrypted via the integrated Cryptography Service (see “Security” above).
Access to client databases is through SOA federation methods.•
. Clients may choose to replace a supplied Data Service with their own Data Service that
uses their databases. As long as the Data Service contract is adhered to, Decision Services
and other application components using the Data Service are unaware of and unaffected
by where the data actually comes from.
. Data Services can be used with off-the-shelf data federation products (often referred to as
“Information as a Service,” or IaaS) for integrating legacy data sources.
www.fico.com page 17
An Architecture for Smarter Decisions
insights »
Access to third-party sources is through FICO data exchange services, a secure, single-point •
of access to a diverse range of data and service providers. The basic exchange infrastructure is
being incorporated into all Decision Management suite applications. Clients select from a menu
of third-party providers those they want to incorporate into their application
decisioning processes.
. The single point of interface is enabled by a Common Data Access Component linking
to some 1,300 networked data sources and service providers for the financial services indus-
try, including credit bureaus and collections agencies. Normalization of data via XML-based
schemas (e.g., credit report, telecom provider data) eliminates the need for clients to be
concerned with each provider’s format and transport method.
. A process engine based on BPEL (Business Process Execution Language) manages requests
for data, including access, parsing and normalization. It supports real-time data access
as well as preconfigured frameworks for batch exchange of job data (e.g., for collections
agency outplacement).
. The Data Exchange is available as an installed service component or as a hosted service
where FICO provides full security and reliability infrastructure. The two implementation
modes have parallel components wrapped in the same SOAP envelopes written in Java.
. In some cases, clients can choose to receive data used for scoring in the form of preloaded
calculated characteristics from FICO’s analytic characteristic libraries. Custom
characteristics built by clients using strategy design tools can also be loaded with third-
party data through FICO data exchange services.
. Templates based on SOA standards such as BPEL, XML and XSLT (Extensible Stylesheet
Language Transformations) are being provided to facilitate the creation of additional third-
party provider “plug in” interfaces.
Common Data Model
The Common Data Model defines the data and data relationships used by the Data Services in
the Data Access Layer. It provides a single logical data model shared by all Decision Management
suite applications. It also comprises the data objects generated from the logical model and their
corresponding database implementations.
A “single source of truth” logical data model spans the customer lifecycle. It enables consistent •
data definitions and unified operational, analytic and reporting databases shared by all Decision
Management suite applications. Common definitions and databases increase efficiency, since
data need be loaded and updated only once for use by multiple applications, and new suite
applications can therefore be brought online quickly. The shared model also supports the
development and deployment of integrated analytics (models that analyze data from more
than one decision area) and reporting that spans the customer lifecycle. It is essential for
connecting decision strategies in originations, customer management, collections and recovery,
and fraud management for better overall results.
www.fico.com page 18
An Architecture for Smarter Decisions
insights »
From this single logical model are generated both the Java classes representing Data Objects •
and the DDL (Data Description Language) scripts for all supported database platforms. Because
these artifacts have one common source, there is 100% consistency across Oracle, Microsoft
SQL Server, DB2 UDB (Linux, Unix, Windows), IBM DB2 for Z/OS (mainframes) DB2 for i (midrange)
database implementations.
Clients can extend the Common Data Model through user-definable tables and/or the addition •
of client XML schemas. A single query can retrieve records across both relational database tables
and XML schemas. Once added, these data model extensions are available to all Decision
Management suite applications.
The Common Data Model integrates with a Software Service that manages text localization •
based on ISO country and language codes. It includes utilities for importing and exporting
localized text via standard interchange formats such as XLIFF (XML Localization Interchange
File Format).
Audit control is provided by a logging Software Service, which audit-stamps selected changes to •
the Common Data Model, such as users updating customer records. A history service keeps
track of the sequence of changes to database entities.
The Insights white paper series provides briefings on research findings and product development directions from FICO. To subscribe, go to www.fico.com/insights.
An Architecture for Smarter Decisions
insights »
For more information US toll-free International email web +1 888 342 6336 +44 (0) 207 940 8718 [email protected] www.fico.com