Friday, August 14, 2009

“Cash for Clunkers” - a Typical Sales Campaign?

monster truck A caller on the radio was commenting that the “Cash for Clunkers” program should have included used cars because the people with the least fuel efficient cars can’t afford a new car - even with a $4500 credit. I don’t know if that statement is true or not, but it got me thinking about the real objective behind the program.

Was the program designed to:

  • get the worst offending gas guzzlers off the streets (environment/ecology)
  • incentivize people who planned to buy a new car, but were waiting for one reason or another to buy now to shore up the Automotive sector and consumer confidence (quick temporary sales increase)
  • provide a unique opportunity to someone who normally could not afford a new car (grow the market)
  • a combination of the above
  • none of the above

One had to meet specific qualifications to take advantage of the program, so was this program designed like any other marketing/sales campaign that targets one or more specific demographics to improve the likelihood of meeting the objective? Should it have been?

Regardless, the advances in technology and availability of data, as a partial owner of the automotive sector, provided the opportunity to use data mining and fact-based analysis to determine the optimum market to target to achieve the program objective. The charts below are an example* of a few types of information that could have been used if this program were treated like a typical sales campaign using data mining to determine who to target and how.

Cluster Descrimination for Demographic Comparison

Naive Bayes Attribute Profiles for Profit Categories

Naive Bayes Attribute Characteristics for High Profit

* The charts are prepared from customer and sales data from a fictional bicycle company, AdventureWorks. The objective of the data-mining in the charts was to determine the characteristics of a customer and their purchase patterns that indicate the probability for a high profit sale.

No comments:

Post a Comment