The Evolution of Cooperative Database Prospecting

When cooperative databases came about in the early 90’s and gave marketers access to a large source of new names, the common selection process utilized a sequential modeling technique where multiple statistical models were used in a specific order to select a universe of prospect names.

This sequential technique enabled marketers to leverage the vast amount of data cooperative databases offered and resulted in huge improvements in prospecting performance over other methods available.

But reaching this new level of prospecting performance created increased complexity for marketers. Although a significant improvement from other methods, ultimately the process left high-performing names on the table.

How so? Imagine that a mailer wanted to generate a 100,000-name prospect universe and used four models in their selection process. Model 1 selects 50,000, model 2 selects 25,000 names, model 3 selects 20,000 names, and model 4 selects 25,000 totaling a universe of 120,000 names.

To reach the 100,000 universe needed, the mailer then chooses to take all of the names from model 1 and 2 and 10,000 names from model 3 and 15,000 names from model 4 leaving a surplus of 10,000 names from model 3 and model 4 respectively.

Were the highest performing names chosen? It’s not very likely that all the best names were chosen and therefore in this example, the 20,000 names left may in fact have been better prospects than 20,000 names that were selected.

Marketers, alternatively, are finding elevated success with a breakthrough approach to selecting prospecting names. One solution to these industry challenges is enhancing the traditional co-op serial modeling approach to identify the best prospects across each individual model. This approach leverages a data-driven algorithm to optimize the prospect universe based on offer attributes, previous buyer attributes, and the multitude of transactional data for each household.

The net result is better and more consistent performance across the prospect universe. In the example above, a mailer can now select a 100,000 name depth from the total 120,000 name universe. The end result is a universe that is comprised of higher potential prospects with an additional 20,000 name surplus that meets the mailers performance criteria.

This next evolution in prospecting fulfillment methodology is resulting in richer prospect universes and higher ROI potential from cooperative database marketing.

Keith Fagan is vice president of product strategy and new initiatives at Abacus Data Services, Epsilon