ICICI Bank : Posidex Dedupe

The credit card issuance process involves evaluating credit worthiness of the
applicant based on a number of market driven parameters and standardized
processes. For this, the credit decisioning system (NACS) interfaces with
various systems and external agencies as part of the decision making process.
Hence a need was felt to allow for a more robust, flexible and dynamic system
which will be parameterized, with no vendor involvement. Integration of NACS
with Posidex for positive dedupe functionality and Blaze Rule Engine (BRE) for
scoring allowed for a dedupe to be much tighter and scoring to be parameterized
and in user control thereby allowing for better control and TAT. Dedupe is an
application matching logic used to catch the applications of the same applicant
who had applied earlier for a credit card.

In this project, the dedupe and Blaze Rule Engine (BRE) databases are hosted
on different servers with GUI options to allow for rapid changes in parameters
to keep up with market dynamics and changes in policy decisions. In the old
dedupe setup, 90% of the applications were having an average of 1-2 matches per
application, while with Posidex, 85% of the applications are having about 1-10
matches per application.

This has resulted in an average dedupe percentage of 47% with Posidex as
against the 21% with the older logic. The total estimated savings will be Rs 225
million PA assuming a 5% increase in approval and decline rates under the
Posidex dedupe match logic and Rs 450 million PA with an assumption of 10%
increase in approval and decline rates. All this has led to an almost 10 times
improvement in processing speeds with minimal hardware requirements.