|
By Tom Groenfeldt
Talk about bad timing. Dealer Services Corporation (DSC) in
Carmel, Indiana set up a business to finance auto dealers' inventory in 2005.
Three years later the financial downturn had not just cut into car sales, it
had put many of their new car dealers out of business as GM and Chrysler dropped
dealers and franchises in bankruptcy. That ended the supply of new cars, and financing,
to hundreds of dealerships across the country.
Many of the dealers, without a source of new cars from two
of the Big Three, turned to used car sales. For DSC, the good news was it has
suddenly lost GM and Chrysler as competitors. The bad news was that its growing
business strained information resources.
For a time, the company used Microsoft Analysis Services,
Excel, and a home-grown ERP system built with Microsoft SQL Server and .NET.
That involved a lot of duplication of effort and data reconciliation. DSC moved
to Information Builders WebFOCUS which proved able to access the company's data
from a variety of data stores and present users with a single clean source of
information.
In used car financing, every time a dealer buys a car for his
inventory he takes out a new loan with its own terms and due dates. Since DSC
provided financing to 9,000 dealers who were supported through 75 branch
offices across the country, this quickly added up to a lot of information. To manage risk, DSC doesn't just track the
loans, it also has to monitor the money it was borrowing to finance all this inventory
and also keep a close eye on the business each dealer was doing.
"Dealers often have bad management skills," said Rado
Kotorov, chief innovation officer at Information Builders. "Loan
originators make the decision on whether to grant the loan and at what interest
rate."
"Our success depends on having accurate information
from our field offices regarding our position with each dealer," said
Chris Brady, DSC's CIO. "With this information in hand, we can predict
each dealer's future financing needs and alert them if they have more inventory
than they are likely to be able to sell in a timely fashion."
WebFOCUS uses the open source R engine for its predictive
analysis, added Kotorov. With it, managers look at the past, present and
future.
"The past is reporting and analysis, the future is
predictive. Now, using real-time information with zero delay, we can perform
the analysis on the data as it flows through the system and before it touches
the database."
WebFOCUS decided to use R for its core analytics engine because
the open source tool has more than 2,000 statistical packages and is supported
by a wide range of universities, research centers and government entities.
"We put services around it including access to data
sources and security." WebFOCUS also provides services around data
cleansing and transformation which, Kotorov said, typically contributes 60 to
90 percent of the cost of modeling.
The state of Michigan has decided to use WebFOCUS in its
foster children program to better match children in need with potential homes
so the children and their foster parents
would stay together longer while avoiding disruption which often leads to
severe problems down the road. With WebFOCUS it built sort of MATCH.COM for the
state's foster parenting program.
"Child welfare is a $7 billion industry," said
Kotorov. "If children change foster parents too quickly, they are likely
to fall out of the system with severe long-term consequences."
The business intelligence engine looks at dozens of factors
for both child and prospective foster parents to determine which group of
parents might provide the best match. A case worker takes it from there and
chooses a home for a child.
Perhaps they will do as well with children as DSC is doing
with its dealers. It has used a
dashboard within WebFOCUS -- the Performance Management Framework (PMF), to break down the trends and their underlying
factors.
"We can drill down to the dealer level to discern who
is most likely to have difficulties, based on past behavior," explained
Brady, the CIO. "This not only reduces our own risk but also helps us
advise dealers so that they can be more profitable."
Brady's team used PMF to create a predictive model that
forecasts future activity based on each dealer's past behavior.
"WebFOCUS
PMF enables us to predict things such as when a dealer will pay off its
vehicles, when they will incur new fees, and how many cars they will be adding
to inventory," she explains. "It is extremely accurate. Our forecasts
are usually within two percentage points of actual revenue."
Kotorov said that WebFOCUS can help improve collaboration
because it gets different groups inside an organization to talk with each
other.
"Typically, the statistical analysts wouldn't have
Boolean analysis skills. It would take them extra time to do data prep. Now the
BI developer can prepare the data and the statistical analyst can run it in the
modeling interface, create the model and save it as meta data. Developers and
BI people don't build end user applications, but once the meta data is on a
server, a developer in BI can build an application on it. This saves time and
the collaborative environment allows everyone to use his or her skills to the
best advantage."
Only registered users can write comments.
Please login or register. |
