Courtesy of ConGA’s Bill Pease:

Here’s my take on various spam scoring systems:

1) Unless your organization’s email content regularly involves
commercial-sounding language (e.g., you are selling products in an
estore, or promoting services provided by affinity marketers),
content-only spam scoring systems are of relatively limited utility.
Content characteristics are generally a minor component of the
commercial or ISP filtering systems that control access to most of your
supporters’ inboxes. Given that, there is not a lot of ROI on the
effort required to tweak messages to avoid the key words or graphical
features that content-only systems identify as “spammy.”

2) Your mail “reputation” is the driving component governing your
deliverability. Reputation is tracked primarily for the IP address(es)
sending your email, not your domain name. Reputation includes whether
your mail infrastructure is compliant with various email standards (such
as sender verification) and whether your IPs are a) sending to large
numbers of invalid addresses at a domain, b) on various blocklists, c)
on commercial whitelists (e.g., Habeas Safelist or SenderScore
Certified) or internal whitelists (e.g., AOL Basic and Enhanced
Whitelists), and d) generating large numbers of spam report complaints
from recipients.

In the last two years, a number of companies have emerged to provide
reputation data in real time to email systems, and you can check at
least aspects of your mail reputation for free at:
Secure Computing’s TrustedSource:
ReturnPath’s Sender Score:
Ironport’s SenderBase:

3) If you want to use a spam scoring system to assess whether a mailing
is going to encounter delivery barriers, you need to use the fee-based
services offered by vendors like Habeas, Return Path or Pivotal
Veracity. In my opinion, PV currently has the most sophisticated
testing tool. In contrast with a free tool like Lyris, PV will check a
message against the 60 most prevalent anti-spam systems, and produce a
report on the full range of message attributes that can impact delivery,
including whether the sending mail system complies with ISP standards or
is listed on blocklists/whitelists, whether domains or URLs within the
message are on blocklists, whether the message content triggers key word
or probabilistic spam filters, and whether similar messages have been
flagged as spam by distributed spam identification networks.

4) Even the best pre-test tool cannot predict your real-world delivery
results, because so many delivery decisions are now made dynamically at
recipient systems in response to real-time data on your mailer IP
behavior and complaint rates. So you should complement any pre-testing
with inbox delivery monitoring. This is also available as a fee-based
service from vendors like Habeas, Return Path or Pivotal Veracity.

Bill Pease, Ph.D.
Chief Scientist
Convio, Inc.
Office 510-540-4772 x2101
Fax 510-540-4163