Firstly, filter-factor is a gross approximation. If you take a photo of a mostly-red scene through a red filter, the change in exposure will be small whereas if you shoot a blue scene through a red filter, the required exposure change is quite large.

A 25A has a filter factor of 3 stops, which is 8x. In theory if you use a TTL meter, it will read the already-corrected exposure but that makes the assumption that the meter and film have the exact same spectral response, which in general they will not. I've found that many meters are quite red-sensitive, so they will report only +2 stops with a 25A whereas that filter has a factor of 3 stops with most films in daylight conditions. If your film lacks red sensitivity (e.g. Acros), then it's even worse. But if the film has extended-red sensitivity (e.g. SFX200), then the filter-factor will be lower!

So you need to be aware of the interaction between lighting spectrum (red/orange filters have little effect under tungsten light because it is already missing a lot of blue), scene spectrum, filter absorption spectrum and film sensitivity spectrum. They all matter, and bundling them into a single number is just a handy oversimplification that works maybe 80% of the time.