Extracting dollars from data

Experian, a unit of London-based GUS PLC, is one of the nation’s largest credit reporting agencies, but it wanted to expand its business beyond credit checks for automobile loans.

If it could collect vehicle data from the nation’s various motor-vehicle departments and blend that with other data, such as change-of-address records, then Experian Automotive could sell the enhanced data to a variety of customers. For example, car dealers could use the data to make sure their inventory matches local buying preferences. And toll collectors could match license plates to addresses to find motorists who sail past toll booths without paying.

But to offer new services, Experian first needed a way to extract, transfer and load data from the 51 different department of motor vehicles (DMV) systems into a single database.

That was a big challenge. “Unlike the credit industry that writes to a common format, the DMVs do not,” says Ken Kauppila, vice-president of IT at Experian Automotive in Costa Mesa, Calif.

Experian didn’t want to replicate the hodgepodge of file formats it inherited when the project began in January 1999 175 formats among 18,000 files and decided to transform and map the data to a common DB2 format.

Fortunately, off-the-shelf software tools for extracting, transforming and loading data (called ETL tools) make it economical to combine very large data repositories. Vendors offering the tools include Evolutionary Technologies International Inc. in Austin, Texas; Embarcadero Technologies Inc. in San Francisco; Oracle Corp.; Informatica Inc. in Redwood City, Calif.; and Sybase Inc. in Dublin, Calif.

Using ETI Extract from Evolutionary Technologies, Experian created a database that can incorporate vehicle information within 48 hours of its entry into any of the nation’s DMV computers.

This is one of the areas in which data management software can excel, says Guy Creese, an analyst at Aberdeen Group Inc. in Boston. “It can simplify the mechanics of multiple data feeds, and it can add to data quality, making fixes possible before errors are propagated to data warehouses,” he says.

New Avenue for Revenue

Using IBM’s DB2 database and the extraction tool, Experian Automotive created a database that processes 175 million transactions per month and has created a variety of new revenue streams. Now, for US$10.99 per query, Experian can make available via the Web the ownership history for any vehicle bought or sold in the U.S.

Car dealerships are a big market for Experian’s database because they’ll pay for data about vehicle ownership preferences in particular geographic areas. Each 17-digit vehicle identification number in the database contains references to model, make and colour. Armed with this data, dealers can determine what kind of vehicle inventory mix might sell best in different regions.

The database which has raised the hackles of privacy advocates includes Experian’s own corporate records, data from 30,000 credit granters and address-change information licensed from the U.S. Postal Service. Plus, Experian is expanding the database to include accident and emission reports, as well as information about vehicle auctions.

The result: Experian offers more comprehensive information than that maintained by state DMVs and auto manufacturers. This information could, for example, help ensure that automakers and auto parts companies are able to contact the majority of vehicle owners affected by recalls even owners who have moved and thereby help save lives and avert vehicular and auto parts-related injuries. Previously, recalls were initiated using dealer service and sales records.

In addition, Experian’s data assets can uncover patterns useful to manufacturers and retailers in creating brand loyalty campaigns and in launching new auto models. Retailers can use the data to speed the process of providing credit to potential buyers. Auto auction companies can check the histories of millions of cars.

Experian’s database is the 10th largest database in the world now with up to 16 billion rows of data. But the company says the database is managed by just three IT professionals, thus demonstrating how efficiently the extraction tools can work with a large database to handle vast amounts of data quickly.

Experian Automotive

Parent company: Experian Information Solutions Inc. in Orange, Calif., a subsidiary of London-based GUS PLC.

Business: Offers the National Vehicle Database, a catalogue of more than 335 million vehicles. Applications include auto history checks, recall notifications and marketing programs. Experian’s North America databases contain more than 65TB of data.

Web site: http://www.automotive.experian.com

Sources: Experian.com, Hoovers.com