How many times have you sent out an important email, then never got a response, only to discover days later from the recipient that it landed in their spam folder? There are two main worries concerning false positives; receiving it, and sending it.

When sending emails, naturally you want to make sure as much as possible that the recipient’s spam filter isn’t going to pick it up by mistake. Since you have no control over what technology your recipients use, you’re left with just a couple techniques—if sending something important, most email clients allow you to request a receipt verification, which is a simple response that recipients click to indicate that they have received your message. Secondly, it’s just a matter of wording it correctly and putting in an appropriate subject line. Common sense rules dictate: Avoid putting in subject lines followed by multiple exclamation points, and avoid spammy-sounding words. Of course, if you’re in the pharmaceuticals business and you have legitimate cause to say “Viagra” in an email, there’s still going to be a risk your emails will be caught by a spam filter as false positives!

Somebody recently hijacked my own personal email address and used it to send out spam, and I received a large quantity of bounceback messages, one of which seemed to be a useful tool. The bounceback message said essentially, that this is the first email message the recipient had seen from my email address, and they are using Sender Address Verification (SAV) as an extra layer of spam prevention. A legitimate sender would see this bounceback, click on a link in the email, and then the email would be directed to its recipient’s inbox.

This technique, sometimes called a callback, does have some limitations. But, in this case, since somebody had hijacked my email address to send the spam, the spam recipient sent the callback (which was sent to my own email inbox). Since I was not the sender, I won’t verify it, and then the recipient is spared from having to view the spam. A limitation of the callback however, is an inevitable delay on receiving legitimate emails that get caught in the callback cycle, since it burdens legitimate senders with an extra step to make sure their emails are received on the other end.

A simple preventive measure is to add all email addresses that you know to be legitimate to your contact list, even if you never actually send out emails to those parties. For example, if you receive email newsletters, adding the return email of the email newsletter to your contact list would ensure that all email from the newsletter publisher would be allowed through the filter. Many email clients automatically add addresses to which you reply to your contact list, but since you do not usually reply to an email newsletter, it doesn’t get added automatically.

The type of spam filter you use naturally makes a big difference, and Bayesian filtering is generally thought of to be the most effective, both in catching spam, and preventing false positives. This type of system will usually not rely just on spotting keywords, a technique which lends itself to a high false positive rate; instead, it looks at the entire message.