MiningSolve™: The Core
Decision Support Tool that Undergirds Our Strategic Approach to
Data Mining Automation.
The
Traditional Approach to Product Cross-Sell or Up-Sell
The traditional approach to cross-sell
and up-sell is through "data mining" the customer database
or datastore. In this conventional approach, a modeler "mines"
the customer information file (CIF) using a manual data mining tool.
The basic approach, regardless of the quantitative technique deployed,
is the same. The modeler attempts to predict the outcome (e.g. customers
who have cross-sold or bought multiple products) using a set of predictors
of that outcome (e.g. other products owned, tenure, demographics,
etc.) The result is an equation, a neural network or a genetic algorithm
that will predict which customers will be most likely to cross-sell
or up-sell. Next, this equation is run against all the customers in
the datamart, producing a cross-sell or up-sell score for each record
in the datamart. There is one score for every product modeled, and
these scores are used to contact customers in a mail-out or calling
campaign.
