For nearly 30 years, fake digital documents have been used to identify external intruders and malicious insider threats. Unfortunately, while fake files hold potential to assist in data theft detection, there is little evidence of their application outside of niche organisations and academic institutions. The barrier to wider adoption appears to be the difficulty in constructing deceptive content. The current generation of solutions principally: (1) use unrealistic random data; (2) output heavily formatted or specialised content, that is difficult to apply to other environments; (3) require users to manually build the content, which is not scalable, or (4) employ an existing production file, which creates a protection paradox. This paper introduces a set of requirements for generating automated fake file content: (1) enticing, (2) realistic, (3) minimise disruption, (4) adaptive, (5) scalable protective coverage, (6) minimise sensitive artefacts and copyright infringement, and (7) contain no distinguishable characteristics. These requirements have been drawn from literature on natural science, magical performances, human deceit, military operations, intrusion detection and previous fake file solutions. These requirements guide the design of an automated fake file content construction system, providing an opportunity for the next generation of solutions to find greater commercial application and widespread adoption.