Inductis Increases Revenue in Information Services
Overview: To increase the accuracy of the information
matching on which its revenues depend, a large
financial information services company turned
to Inductis to create a new, state-of-the-art
database search engine.
Inductis created a powerful information
matching system that enables real time searching of millions of records to produce
relevant matches from fragmentary information. With its sophisticated algorithms,
scalability, and flexibility, the system has wide application in such information-rich
areas as financial services, telecommunications, biometrics, and life sciences.
Situation
The revenue of a large financial
information services firm depends on the speed and relevance of its information
search system. Clients submit fragmentary information about companies they want
to identify and about which they wish to purchase a report. The database search
engine of the financial services company attempts to match the fragmentary information
with the correct company from among some 30 million possibilities. Because the
firm earns fees as a result of accurate matches only, even small fluctuations
in the system's error rate can have an enormous impact on revenue, with each percentage
point change representing millions of dollars in revenue.
The company's existing
system was generally efficient, but its mainframe-based application for performing
business matching was running at an error rate of 10 percent. As a result, the
company was losing revenue as well as market share. To help address the problem,
the company turned to Inductis.
Developing
the Matching System
Leveraging
cross-functional capabilities in business strategy, analytics, and technology,
Inductis managed a team composed of consultants and client personnel to address
the problem. Recognizing that the challenge was essentially to build a better
database search engine, the team established ambitious goals for a new system:
A match rate 5-8% higher than the company's existing best-in-class match system
Peak throughput of 550,000 transactions per hour
A response time of less than 0.8 seconds
24
x 6.5 availability
Scalability
to support yearly growth rates of 10%
Interoperability
with the existing online system and with future systems
Easy
performance reporting and monitoring
The
project was designed to enable parallel development, testing, and integration
of the new system and to deliver it on budget and in a compressed time frame.
The team identified the requirements of the new algorithms that would power the
database search engine, designed them, and prototyped them. Simultaneously, the
team developed a three-layered architecture with redundant and scalable UNIX servers
at each layer to provide a reliable solution that would meet business requirements
and replace the original mainframe system. The new system was then integrated
into the existing system and designed to support a Web services interface.
Unparalleled
Performance
Using the latest
technology such as JAVA, Message Queues, and in-memory data store, the system
now provides the client with a best-in-class search capability. Within 800 milliseconds,
the Inductis-built system can rummage through 30 million records and over 100
gigabytes of data and identify precisely the right company. It has improved match
rates by over 5% and provides two to three times the throughput of the existing
system. Having met or exceeded all of its initial performance goals, the matching
system is being rolled out to all of the company's online products.
Results
Results to date have been impressive. The
company has achieved a steady 5% increase in company-wide revenues. In its first
year of operation, the match system will produce $30-50 million in net benefits,
including a PTI impact of $30 million. With new-found speed and accuracy in its
search capability, the company is well positioned to regain market share and to
scale up easily to meet large increases in volume in the future.
