We analyze empirical data of Germany’s largest online social lending platform Smava.de to exemplarily study the contribution of unstructured, ambiguous, or unverified information to trust building in online communities. After controlling for the influence of hard information, we find that textual statements that appeal to social behavior actually affect trust building. However, the evidence is less clear for voluntarily disclosed personal data. Lenders generally seem to give more weight to hard information so that disclosing personal data promises little benefit while potentially exposing borrowers to privacy risks. Key words: Trust Building, Soft Information, Online Social Lending, Social Behavior, Reciprocity, Personal Data, Privacy, Empirical Study, Content Analysis