Business applications of data mining in marketing often focus on use of predictive models to classify customer events such as acquisition, sales of products and services or customer churn. These models frequently utilize internal, individual customer-level records, as well as external, socio-demographic information, available only at postal-codes or some other level of geo-spatial aggregation. In order to efficiently acquire new customers companies target individual households using unaddressed direct mailings distributed to targeted geographical areas. In our poster we examine how data mining help identify best geographic areas for customer acquisition campaigns. In addition we study the effect of data aggregation on measuring model's performance. We suggest that population-based model performance measures are used rather than geo-spatial based measures. The work is illustrated with an actual acquisition campaign data from the field of telecommunication in Canada. KEYWORDS Marke...