Systems and techniques to generate statistical reports on transactions conducted via an electronic marketplace are based on data extracted from the transaction documents. In general, in one implementation, the technique includes receiving documents sent through an electronic marketplace, and extracting data from the documents. The extracted data may relate to a predetermined statistical category of transactions conducted through the electronic marketplace. The extracted data may be stored for each document, and the stored data may be aggregated according to the predetermined statistical category. In some implementations, a report may be generated that relates to a document flow analysis for documents sent through the marketplace according to the buyer, seller, time period, document type, and/or other parameters.