A data validation architecture is provided that minimizes duplicate code and enables easy creation of new validations using a high-performance and scalable design. Such an architecture provides sufficient flexibility to disable certain validations when they are not applicable to the flow, interpret validation failures differently according to requirements of the calling application, and specify the type of document for which a particular validation is applicable. The architecture provides a standardized flow for performing data validation, specifying a set of required parameters for validation subroutines, as well as where validation results are to be stored for access by a calling program. Such an architecture also provides a framework for performing common processing required for most validation flows without restricting the flexibility of the validation subroutines.