How spam filters score emails
Barracuda Spam Firewall is an expensive hardware spam filter that is installed by large organizations within their own datacentres.
Barracuda reports a single score, between 0 and 10 and a flag indicating if the email is considered to be spam. You should be aiming for a score below 3.5.
Please note, for some very spam-like emails, Barracuda will quarantine the email. In these cases, Litmus will report the email as considered spam and mark it with a score of 10.0 (the maximum).
MessageLabs is an expensive corporate server-side spam filter. It is in use by many large organizations and is very regularly updated.
Somewhat confusingly, MessageLabs score in two different ways. On most emails, MessageLabs will actually report the score from a very well configured SpamAssassin installation, but there is no threshold in this case. So even with a score a 4, the email may be marked as spam.
Sometimes, MessageLabs will process the email using its own filter, which reports a very similar score to SpamAssassin but rounded to the nearest whole number. In both cases, a score of 3 or more is to be avoided, but Litmus will always report if the email was considered spam by MessageLabs as a separate calculation to the score.
SpamAssassin is a very popular open source spam filter. Here we are testing against version 3.3, using its default settings.
SpamAssassin comes with a large set of rules which are applied to determine whether an email is spam or not. The scores can be positive or negative, with positive values indicating spam/failing emails and negative values indicating passing emails. The higher the positive score is for your email, the higher the probability that the message is spam.
Generally, your email should have a score of 5.0 or lower to be considered passing. The lower your score, the more likely your email is going to be received in your subscribers' inboxes.
Outlook utilizes a self-learning filter to determine what you think is spam. While this is great for individual users, it's not consistent nor reliable for use across thousands of tests on our servers. Instead, we've added in hundreds of spam rules that have been published by Outlook. Whenever the content in your campaign triggers one of these rules, we'll provide you with feedback on what can be changed to make your email look less "spammy" to Outlook.
This filter on Litmus uses built-in junk email filter for Outlook, which ships as part of Microsoft Office. This has various sensitivity settings, here we have set it to 'High'. The Microsoft Outlook filter scores from 0-10 on the High sensitivity rating, with 0 being the highest (passing) and 10 being the lowest (failing). Outlook rates an email with a 6.0 or higher (out of 10) as a failure. A lower score (lower than 6.0) is considered a passing score with the High sensitivity rating.
Emails that have successfully been received in the email will show as a "passed" score. Anything that has a "failed" score will indicate that the email was not received in the inbox, which means the email could have been blocked or sent to the junk folder for that particular email client. You will also want to make sure that you are sending in your email to the complete list of seed list addresses in order to see the accurate reputation scoring for your email.
Part of the issue of testing for spam on webmail providers (such as Gmail, Yahoo! Mail, Outlook.com, and AOL Mail) is that a large portion of their spam scoring is based on internal metrics and algorithms -- for example, it's common for a single email to send to a single inbox on Yahoo! to be delivered successfully, but that same email may become suspicious when sent to ten or hundreds of thousands of recipients. Sometimes, a portion of recipients complain or do not engage with the email, leading the rest of the send to be delayed or marked as spam. Trends in engagement from a particular sender or IP may also produce blocks or delays, as do spam filters that "learn" about what individual users consider spam through open, click and reporting behaviors. These scenarios are difficult, if not impossible, to predict.
The webmail filters do not provide us any further information as to why an email has not been received in the inbox. You may want to check warnings produced by server-side filters (like Barracuda or SpamAssassin) and update your email with the changes those filters have suggested. You may also want to check with your Email Service Provider to see if users for a particular webmail service have marked your previous campaigns as spam in the past, as this can affect your spam performance for your email campaigns sent now.
See how your emails stack up
Now that you know how the scoring works for each spam filter, it's time to try it for yourself. How do your emails stack up?