How Does Content Filtering Work?
Aug 29, 2007
In his ClickZ column, Stefan Pollard gives us some insights into how content filters work:
"Many major content filters rate messages according to how many elements in them are commonly associated with spam, such as copy, format, coding, and headers.
A content filtering system doesn't automatically flag a message because it contains one or two words associated with spam. Instead, a system like SpamAssassin runs hundreds of tests on a message, searching for spam-like elements. When the system encounters one, it assigns a point value, from a fraction for a suspect word to several points for major infractions, such as being blacklisted.
If a message accumulates too many points, the filtering system flags it as spam. Just how many points it takes to cross the threshold depends on how the filters are configured at the destination. Some are set very low, with three to five points being all that's needed. Others are more liberal.
More issues are also likely to trigger filters and collect higher point values than those assigned to words in the message body. They include:
- Incorrectly formatted or incomplete e-mail headers, which list technical details of the message transfer, including the sender's IP and e-mail sending address
- Broken tags and sloppy HTML coding
- Too large an image relative to the amount of text
- Scripting that could launch viruses or spyware
- Attachments
- URLs or domains that have appeared in spam
Using "free" in a subject line or "click here" in the body may add a fraction of a point to a total content score, but if other e-mail elements are in good order, the message shouldn't be flagged as spam.
How concerned should you be about content filtering?
Most of the time, major portals such as Hotmail, Yahoo, and AOL rely more on the sender's reputation than on message content. Even if content scores high with the filters, a good reputation should help messages get to an inbox.
Most corporations rely on content filters to block spam. Thus, content plays a larger role in business-to-business (B2B) message delivery.
You know your market segments. Analyze domain names in your e-mail address database, and you'll see your major destinations."
Source: ClickZ