Can SpamAssassin be trained from Outlook?

I'm comparing and contrasting SpamNet from Cloudmark, Spam Bully, and SpamAssassin. The first two are personal spam filters that integrate into Outlook, and the third one is of course the anti-spam software that can be turned on or off on a per hosting account basis within cPanel.

Between SpamNet and Spam Bully, I marginally prefer Spam Bully. It seems to have a slightly higher success rate in determining correctly what is and isn't spam.

But seeing as we host about 150 websites for our clients, we are looking into the practicalities of recommending (or not) that they turn on SpamAssassin on their hosting accounts.

The biggest problem I have with SpamAssassin is this: as far as I can tell, SpamAssassin cannot be 'trained', via Outlook, and on a per-pop account basis, what is and isn't spam. So if SpamAssassin decides that something is spam, then it will forever tag it as such - and there's nothing the recipient can do about this. Sure, a savvy representative of the company could go into cPanel and manage the Spam Assassin whitelists and blacklists, but what would be infinitely more preferable would be a plug-in for Outlook (just like SpamNet and Spam Bully) that allows each individual POP account holder to indicate what is and isn't spam, and for this information to be instantly communicated back to the server.

In summary - until SpamAssassin can be trained by individual POP account holders from Outlook, then I can't ever see that I'm going to recommend it to my clients as an acceptable solution over SpamNet or Spam Bully.

Comments anyone?

 

I had a go with Outlook 2003's junk email feature. Not bad, but not (IMHO) as good as Spam Bully which seems to be a little cleverer (better bayesian filtering I guess). So I turned off the junk email filter and turned on Spam Bully.

That said, I'm quite sure that over time Microsoft's junk email filter will get better, and given that it's built-in, ready-to-go in Outlook 2003, I'm sure it will blow away the competition who won't be able to compete with a half-decent and improving built-in free product.