Hi Ramorse,
Spam can be difficult to tackle as it is ever changing and classification can differ by user. SpamAssassin is quite a powerful tool, but I too have seen more people struggle with it lately and often times not taking much time to customize to their needs. SpamAssassin has a generic
guide for this. Additionally, upgrading to 11.46 will provide an update to SpamAssassin.
Most are aware of adjusting
thresholds and
customizing SpamAssassin.
Many also suggest to enable
bayesian filtering in SpamAssassin which requires 200 ham and 200 spam tokens to even take effect. I find this quite effective; however, it can be very difficult to get users to place spam in the spam folder and ham in a ham folder. Another fault with bayesian filtering is the possibility of bayesian poisoning. I personally think this is a large part of SpamAssassin's issues of late. Cleaning compromised mail servers, I am seeing more spammers include large text from books scraped from sources such as Google Books as a method to legitimize the message and to smudge the line between spam and ham.
Another method is trying new
rulesets. Be careful here though, as using an outdated ruleset(SARE) can cause worse accuracy.
SpamAssassin also has the
Pyzor and
Razor plugins, but I can not offer any information on these as I have no personal experience with them, yet.
Now to end my rant and address your actual question, I think a simple and effective solution is to use
RBL's. I am a fan of SORB's myself. I can't provide any information on the 3rd-party you've referenced, but if you're looking to outsource your mail, I'd personally suggest trying Google Apps mail. Google has spent tons of money on new methods of identifying spam effectively, but frankly, it comes down to the volume of mail they handle, which gives them a considerable edge.
Thanks,