Anti-spam is a topic on which I have strong feelings.
It is no failing that SmarterMail is not an anti-spam solution. The "post office" function and the anti-spam function are very different product spaces. For instance, anti-spam needs a daily feed of a large and diverse mail flow. SmarterTools is not in that business.
Anti-spam should run on an incoming gateway acting as your MX, so it is independent of your "post office" choice. There are many options, both cloud and on-premises, that can be placed in front of SmarterMail. The challenge is find a competent product.
The purpose of an anti-spam product is to block unwanted mail and allow wanted mail. That involves some guesswork and some personal preference. Therefore, the measure of a good anti-spam solution is not what it knows, but rather whether it gives you the tools you need to define wanted and unwanted mail to a high degree of precision. It is no flaw that Kaspersky made judgement errors; the flaw is your implication that it could not be tailored to easily correct those errors.
"Unwanted" messages can be broken down into these groups:
- Malicious messages that nobody wants. The commercial products should be best at this, although I have my doubts that they actually deliver that much.
- Messages that are unwanted based on company policy. Your mail system administrators will be best at this, and the filtering system needs to be able to reflect their wishes.
- Messages that are unwanted by the recipient. The person is the best judge of that question, and the "post office" system needs to provide a way for the user to provide that feedback.
SmarterMail has a wonderful solution for user feedback, called the Training folder. When enabled, anything that a user places into their "Junk Mail" folder is copied to a folder on the server, then removed after an hour. A couple of integrated products can pick up those files and use it to feed their Bayesian database. I don't believe in guesswork and I don't have those products. I copy the message to a folder for review each one manually, deciding whether to block the message source, unsubscribe the user, or ignore the complaint.
My daily routine involves:
- Checking the quarantine folder for items that should be released. Filtering rules are adjusted so that the next message with those characteristics will be allowed.
- Checking the list of messages from unknown senders, for allowed messages that should have been blocked. Filtering rules are adjusted so that future messages will be blocked, and particularly dangerous messages are removed from the user's mailbox by the impersonation feature available to system managers.
- Checking the user feedback folder to adjust filtering rules to accommodate user preferences.
I have been very disappointed with the anti-spam products that I have surveyed, and I have looked at a lot. I still use one commercial product for content filtering, but most of my filtering is based on sender identity, not content. Maybe some people think you can give big money to Mimecast or ProofPoint, then forget about your spam problems forever. I don't.