E-Mail Bounce Management Using Text Mining

Direct marketers assign the communication channel e-mail an increasing importance, evolving a marketer´s collection of customer e-mail addresses to an asset of important business value in order to provide customers the right offer at the right point in time. Besides the design and content of an e-mail campaign the e-mail-transportation process is of critical importance to really reach your potential customer. Possible errors in the transportation can be of either permanent - hardbounces - or temporary nature - softbounces. With the transportation error communication by Internet-service-providers lacking of compliance to existing standards, companies face the problem of having inaccurate information regarding deliverability. And quite often this results in the blocking of actually intact potential customer addresses.Therefore a model processing bounce messages and predicting the possibility of deliverability using semantic information obtained through text mining and the embedded singular value decomposition is presented. We discuss the application of text mining tools regarding error messages and apply a multi-staged decision model that minimizes the false-negative-rate to discard addresses with permanent non-deliverability from the database while identifying addresses that maintain business value although being bounced.