Making sure the structure of the SFM file reflects what the linguist intended is crucial to doing an import that gives the linguist the closest thing to what they last had in their Toolbox file. Importing the wrong structure really constitutes "loss of data", because they are having to fix things that didn't seem broken to them when they were in Toolbox. (They are aware of all the lexicography they need to continue to do; they don't want additional work added on to fix things that are the result of the import process.
Below are some specific steps for validating the structure. In the process, you may discover places where you need to "make a best guess" about what was intended in the data. However, it is important for the linguist to be able to find and review these before he begins his lexicography work with the FLEx project you send them. There needs to be a way to mark these places.
Before doing further work on adjusting the database for import, it is important to validate its structure using Solid.
There is a page that is a start at describing some principles for how to work with Solid, and there are videos about Solid on the DLS Lexicography Course site.
In general, the kinds of things you are trying to accomplish are:
Make sure that Sense level information is in senses and Entry level information is in entries.
Move any Entry level fields that occur in or between senses, to a position that is more clearly an Entry level position.
Change any markers that have been used at more than one level, so that each marker is only used at one level (when that matters).
Determine whether subentries occur only on the Entry level, or both Entry and Sense (and make needed adjustments).
Make sure fields that should only occur once at their level indeed occur only once.