Is there any reason why a file analysis using a cloud engine returns zero matches?

Here is our scenario at our place of work:

1. We have Trados Team that we used mostly to store and query our resources (TMs and terminology databases), we also have there our MT provider of choice connected (via API), and we have also our engines for each language pair that we work with. Those engines include the TMs, TBs, and MT provider mentioned.

2. Each translator and our project manager all have a stand-alone version of Studio Trados 2022. Projects are created and worked on locally. The online editor is not used because it lacks all the functionality we are used to.

3. Translation projects are created locally using the appropriate engine. The file analysis returns zero matches in every instances. We work in an environment where there is a lot of recycled material. So we are absolutely certain that there must be TM matches.

4. If the same project is created also locally and using exactly the same resources locally (exact copies of TMs and TBs, plus the MT provider), the analysis feature works as it is supposed to work. An 800-word document may turn out to be only 300 words new words, while the rest is an exact matches or fuzzy matches.

Any reason why this is happening?

Your input will be greatly appreciated.