This article centers on characterizing patterns of language use in online auction sites (the ‘e-marketplace’), with a particular focus on the language employed in product descriptions listed on eBay. Via a corpus-based assay of a sub-corpus of a 6.3-million-word corpus of eBay data, the authors try to gain a better grasp of variance in descriptions, depending on two vital factors: seller experience and item condition (a high number of ‘new’ listings for a specific seller is likely to be a sign of a more experienced seller). This article describes the methods and approaches employed to construct the eBay corpus and supplies a detailed assay of a sub-corpus of data as a means of distinguishing some of the pivotal transactional and interactional features found in this genre of discourse.