A user-driven document data collection system may allow the user to enter document data in no particular order. The data collection system may help the user identify documents and determine whether those documents are relevant. The document data collection system may also allow the user to input a description of a document, identify the document based on the description, and determine whether or not the document is appropriate for data collection. The data collection system may be configured to display example documents for the user to verify the identification of a document. A user may enter data for a document via a data entry screen based in part on a scanned image of the document. The document data collection system may analyze the data from documents to determine whether or not any additional information, such as from additional documents, is required to perform a particular task using the document data.