Verification, Validation and Accreditation/Certification (VV&A/C) have long been required steps in producing a Model & Simulation (M&S) that will satisfy the user. It has also been largely under-resourced, sidestepped or ignored because of time and cost constraints. One of the perceived difficulties in performing VV&A is the certification of the data used in M&S systems. Data collection often commences with little or no definition of "what specific" data needs to be collected or how this data will be used in the resulting software model. Interjecting verification and validation of data, and hence data certification, at appropriate points during software development, can occur with little impact on development schedules. To be sure, certification of data is not free, but it does not have to be difficult and it can be accomplished within reasonable resource constraints. More importantly, this certification is absolutely necessary to achieve a valid model that can then be accredited for its particular use. This paper will show when data verification, validation and certification should occur during system software development. Additionally, it will propose a redefinition of data certification that will limit data use based on the validation of the data and the model that will use the data.