Elisabeth Orion 

Are you aware of the limitations here?

https://support.deepl.com/hc/en-us/articles/360021634540-About-the-glossary

Glossaries created via the DeepL website and DeepL desktop apps are different to the glossaries within your CAT tool. This means that these glossaries cannot be used within your CAT tool and vice versa.

Given your test comparison showed it working in the browser I suspect it's because you are using a glossary created in DeepL and not created via the Trados Studio DeepL plugin interface.  Can you confirm that?

I don't think so and I just checked back with my colleagues.
We have a separate software to manage our terminology (Termbank), we export our glossaries from there and then upload the same glossary to DeepL and Trados as a csv. The respective glossaries and DeepL-MT are activated by project management when creating a new project in Trados. 

Glossary terms are almost always used correctly, but we have noticed a few instances where they aren't used, most noticably the spelling mistakes with z as described above.

 Elisabeth Orion 

Thanks for that clarification.  Shame such a process is needed, but I guess we need to wait for DeepL to support glossaries in a more effective manner.

In the meantime, you mention this is almost always used correctly, but sometimes it's not.  That makes it tricky and I think the first thing you need to do is demonstrate where it's not working by sharing an example of something we can reproduce.  So if you can provide the following small examples:

1. sample source file
2. sample csv

Then we can test this ourselves specifically.  If it helps to reproduce the issue then a developer will almost certainly fix it.

You also seem technical given your reference to the API documentation so you might also be interested to look at the source code too as this is an open sourced application: https://github.com/RWS/Sdl-Community/tree/master/DeepL%20Translation%20Provider

The appstore team will welcome any pull requests from 3rd party developers who find a problem and fix it.  Often a faster way since we only need to validate the code and publish the fix.

 Elisabeth Orion 

Thanks for that clarification.  Shame such a process is needed, but I guess we need to wait for DeepL to support glossaries in a more effective manner.

In the meantime, you mention this is almost always used correctly, but sometimes it's not.  That makes it tricky and I think the first thing you need to do is demonstrate where it's not working by sharing an example of something we can reproduce.  So if you can provide the following small examples:

1. sample source file
2. sample csv

Then we can test this ourselves specifically.  If it helps to reproduce the issue then a developer will almost certainly fix it.

You also seem technical given your reference to the API documentation so you might also be interested to look at the source code too as this is an open sourced application: https://github.com/RWS/Sdl-Community/tree/master/DeepL%20Translation%20Provider

The appstore team will welcome any pull requests from 3rd party developers who find a problem and fix it.  Often a faster way since we only need to validate the code and publish the fix.

Thank you for you messages - I will try to gather examples and take a look at the github repo. I can hopefully get back to you soon.