During one of my presentations at DMA06 in San Francisco last week, an altogether common question was raised: Can words like "free" cause email messages to be unduly filtered as spam?
In my experience, content filters don't affect deliverability as much as they used to.
In the early days of spam filtering, increasingly intelligent content filters were used to look for patterns that could predict whether email messages were spam. Words like "free" or "spam" or "XXX" would flag a message and block its delivery. And, while this is still true today, the content part of spam filtering has taken a back seat to the more accurate and relevant reputation metrics.
Why? Because no matter how smart programmers made their content filters, human spammers could always find a way around them. This cat-and-mouse game shifted the focus toward IP-based reputation systems. This meant that ISPs and spam filters started paying far more attention to users' "spam button" complaints, thus pushing the spam filter function back onto humans rather than computers. And, over the years, this approach has proven effective. The problem of spam in the inbox has been alleviated, and users' perception of spam has softened somewhat.
My view on this was bolstered by a recent study released by ReturnPath that placed a specific metric on the weight of reputation vs. content. The figure ReturnPath came up with was 83 percent. You can read more about it in this article.
Even if content filters today only account for 17 percent of a spam score, they still matter, and can't be ignored. That's why companies like mine include built-in content checking systems that you can use to test your messages before they go out into the real world.