If you use Lightroom this is my workflow -
I use Lightroom Classic, which does an initial metadata extraction into its own catalog when files are imported, and treats the media file as an object regardless of whether the underlying file is there or not.
This means I can import a video into Lightroom, which will read and keep the original capture date, GPS coordinates, and keywords I’ve added - and then I do things like convert to h265 or trim in a separate app - and then tell Lightroom that the catalog asset now points to (for example) 1234-trimmed.mp4 instead of 1234.mov.
For me this works because my workflow is then to use a plugin to automatically publish changed media files to my PhotoStructure media folders (not finding PS reliable at picking up changes/delete to existing files, I now do regular PS library rebuilds which take a day or two 
).
This workflow wouldn’t work if you pointed PhotoStructure directly at your Lightroom media folders as the metadata would be sitting in the Lightroom catalog unwritten to the file. There are ways around this, such as using an exiftool writer for Lightroom.
And if you don’t use Lightroom… Well ignore everything I said 
Happy to answer any questions if useful.