I've elaborated on this in other posts, so I'll give just a brief recap here.
I, too, apply factors after metering through the filter. These, however, are not "Hutchings' Factors," rather a compensation for the changed response of different films to different colors of light. (FWIW, the logic behind the Hutchings' Factors seems flawed to me; a red filter should keep blue light from reaching the meter cell and give a correspondingly lower reading, which would then be placed appropriately to retain the desired detail.)
From what I understand, and from what I've learned from empirical testing, films respond differently in regards to contrast and sensitivity to strongly-colored light as opposed to white light. This seems to be dependent on the sensitivity characteristics of different films, probably due to the different ways of sensitizing panchromatic film to the green and red portions of the spectrum. One film's red-sensitive portion may be contrastier and slower than the film as a whole, for example.
Strong filters such as dark red, dark green and dark blue limit the spectrum of light reaching the film the most, and these are the filters that seem to require the exposure/development factors. I test the strongest filters with the films I use to arrive at these "fudge factors," which include both exposure and development compensation.
I find that not all films behave the same way. Tri-X, for example, needs additional exposure and reduced development with a #25 red filter (for me, + 2/3 stop and N-1). TMY, on the other hand, needs more development in addition to extra exposure to reach the same contrast. With dark green filters, TMY needs less development, but Tri-X needs more...
Fortunately, testing is fairly easy and needs to be done for only the strongest filters one uses (I didn't find appreciable difference with #8, and #11 filters, for example to warrant a factor). Most of us don't use dark blue or cyan (and if so, then less often), so the dark red, orange and green filters are the most important and the ones we can start with.
Test neutral subjects with and without the filter and compare exposure and contrast and adjust exposure and development for the filter so that these parameters match (very similar to Zone System calibration). Neutral subjects are necessary, so that additional changes in tone are not introduced into the tests by the filter's transmission of different colors.
Once you have the factor for each filter and the films you use, you're good to go.
This refinement makes metering through the filter a really reliable and useful tool.