The simplest possibility to obtain the plasma frequency is to model reflectance or the result of ellipsometric measurements with the help of the Drude Model. Transmittance is usually hard to measure below the the plasma frequency since the reflectance is very high then, but if there are no disturbing interband transitions then you might be able to see a strong increase in the transmittance above the plasma frequency which is the stronger the higher the plasmon lifetime is.
With a bulk metal (I assume you are thinking about metals) showing a flat surface, you can do a direct inversion of spectroscopic ellipsometry data to get the "pseudo" dielectric function of the material. It should be the same whatever the angle of incidence. From this dielectric function you can get the screened plasma energy.
For a thin film, first you have to make sure that its structure is not granular/discontinuous, otherwise it will not make sense to define a "reflection edge". Usually this happens for thin layers of metals. How thick is the film? What is its composition? It can help to examine the reflectance and transmittance spectra. Does the film show a 0% transmittance at visible wavelengths? What is the shape of the transmittance spectrum?
How did you calculate alpha, hopefully not from -lgT = A? I would strongly advise against employing the Beer-Lambert. Hopefully you used the Abeles formalism in order to fit a proper dielectric function model. As I wrote above from that you automatically get the plasma frequency in a proper way, see e.g.
Article Comparison of ion-beam-assisted molecular beam epitaxy with ...
Article Employing Theories Far beyond Their Limits-The Case of the (...