Hello,
1) Inspect the mail headers mails. If Bayes learning is working, you should see a BAYES_SPAM tag with a higher score. It’s possible the filter is learning correctly, but the message simply isn’t reaching the spam threshold yet.
2) Make sure your IMAP client isn’t creating its own “Spam” or “Junk” folder. If it does, your messages may not be moved into the folder Mailcow actually monitors for learning, which means nothing is being trained.
3) Also Bayes need certain number of messages to be able, so if you mark only few messages you need to continue,..
4) Maybe check Rspamd logs to see more information, you should be able to see there if it’s learning or not,..