Irrespective of the use of Sketchup or not, any method you can find to remove unwanted entities can potentially save unecessary polygons being drawn.
One important way it can help is by removing polygons from areas FPSC draws them when the player is nowhere near them or when they are not part of the scene to view often in a part of the level the opposite side of a level to the player or already in an area the player has already gone past in a linear gameplay fashion and so there can be many instances where polys could be removed.
Dont forget that FPSC can unecessarily draw polys of segments as well as entities in this way as part of its compile process somewhat erronously or perhaps not but in any case they are often of no value being nowhere in camera view.
Thus one can even convert some world object segments to entities and Load and Unload them. Looking in wireframe in editor 3D view will show you if FPSC is creating segments far away from player out of sight eating away at fps. If you can convert any to decoration as entities instead you can force their removal when FPSC may otherwise draw them.
Dont forget there is a Law of diminishing returns here to consider. You can save fps up to a point until the strain on the engine of having to track what it has to do in loading and unloading and actually doing it on the fly will balance out any benefit.
Finally you can optimise and save fps, however unless you can successfully do it with Characters and other Dynamic AI entities you wont save maximum fps - their AI burden quickly kills fps much more quickly than large numbers of polys do.
FPSC is considerably less efficient in handling AI entities now as far as I can tell than it was when I created the Load and Unload Entities on the Fly scripts and removing them wherever possible when not needed is now more vital than ever.
Examples of things you can successfully remove apart from Characters are, Barrels, Boxes, Crates, Grills, Caves, steel roof girders and almost any kind of decoration and even parts of buildings or whole chunks of a level if you do so carefully.
Optimising. Its time consuming to do. Is it worth it?
Thats for others to decide.
Maybe one day when FPSC is more stable in areas of concern where fps is unduely influenced by its own circumstances, then such a feature as Loading and Unloading of Game objects of many kinds could be successfuly used to fully give back maximum fps to users during gameplay even when large numbers of AI entities are present to view. Then it would be worth developing the potential further.
This time next year perhaps.