Authoring rules for filenames and data is not enough, for a project, you need to extrapolate names as soon as possible to make sure that your name strategy works. I also get fed up when industry says “we don’t know this early”. How did you tender the work then? Guess???
I have seen this far too many times. Not enough thought given to exploring and projecting naming things. We issue references to 1192 etc but these are left then to interpretation of the “rules”. It is much more straightforward to “model” the names of things that are know at the beginning. Reports, minutes, agendas, models, federated models and then actually create placeholders in the system that is going to be hosting these entities. Be that online or local. Why? You can track progress for a start. You can set dates for when these objects should and will be updated. You can actually get people to understand and get to grips with what these names actually mean from the start. Not just hand out a set of rules and expect/assume people will get it write first time. Mapping out as much as you can (which will be more than you think) right back at the beginning. We used to call this a storyboard or cartoon set. The practice has died out, bring it back, it was useful!