How We Label Spam
The PureMessage software analyzes each message, using a large set of subtle rules to determine the likelihood that the message is spam. If the spam probability adds up to 50 percent or greater, then the message is flagged as spam in two ways.
Flag #1: "[PMX:#]" added to the Subject line
The subject of a spam-flagged message will look like this:
Subject: [PMX:####] BID NOW On These Seized Goods!
Each # represents a 10 percent increase in the probability that the message is spam. Since we start flagging at values above 50%, one # indicates a 51-60% probability of spam, while ### means the probability of spam is close to 80%. You'll never see four or five asterisks, because we automatically reject all messages that score that high.
Flag #2: "X-Perlmx-Spam" header added to the message
The X-Perlmx-Spam: header shows in detail why the message is considered to be spam. Here is an example:
X-Perlmx-Spam: Gauge=XXXXXXXIIIIIIII, Probability=78%, Report="CLICK_BELOW, FROM_ENDS_IN_NUMS, HTML_FONT_COLOR_GRAY, HTML_FONT_COLOR_UNSAFE, IMGSPAM4, MAILTO_LINK, etc., etc., etc."
The Gauge shows the probability that the message is spam; the Report lists the clues that PureMessage used to calculate the spam probability.
What you can do: Set up e-mail filters to watch for these flags
In your e-mail program, you can set up filters to watch for "PMX" in the Subject line of incoming messages. We recommend that you filter spam into a separate mailbox (calling it "spam" is ok) and then peruse the mailbox occasionally to see if messages that you consider not to be spam have landed there; then you can adjust your filters as needed.