Mark "Not Spam" = Positive IP Reputation?
So I'm an avid user of Netflix. I usually go through about 6 DVDs every two weeks or so. Their email notifications are exceptional in their on-time delivery and are always accurate to the order of my queue. These message always land in my Yahoo! inbox without fail.
This past week, they've seemingly had some issues on their side as notifications were not released upon the return of a specific DVD and with which DVD you should expect to receive next. There's a message on their main page explaining the issue and they also emailed their entire database about the issue. But unlike their notification emails, this announcement landed in my spam folder. Now I have a theory as to why this happened: They sent one mass email to their entire database at once explaining the issue, thus making email clients think it was a spam message due to the large amount of recipients it was going to.
But from this debacle, it dawned on me that if a spammy message lands in your inbox, you usually mark it as spam and it consequently effects the IP address negatively. But what if you mark it as "not spam"? Don't you think it should inversely effect the IP in a positive manner?
After discussing this with a few co-workers, some interesting points were made. With this scenario, there's a strong possibility that spammers would just have to create multiple accounts across all major platforms to counteract the legit spam reports, thus increasing the deliverable spam email. But what if major email clients could track the "not-spam" marks on a person by person basis? I'm not sure how likely that is, but if someone figures it out, it's definitely something I'll support...
Being involved with email, deliverability is one of my biggest challenges. It's a constant struggle to combat spam filters. But if something like this is initiated, it would definitely help my cause and make my job a lot easier.
Does anyone know if something like this is feasible? Do you think it's a good idea? Bad idea? Neutral? Comment below!
Labels: Email Deliverability, IP Reputation, Not Spam
posted by MindComet at 3:57 PM
3 Comments:
The "Not Spam" button is one of the key indicators that ISPs use to check their filters for false positives, it is also a major indicator for messages that made it past their filters (a'la "report spam" button).
One likely cause of the bulk delivery is that the message sent was so different from their normal communications that the systems confused the content with spam and thus filtered the messages. Likely the regular mailings are tested and coded differently to work in a positive way with filters, where the kind of testing/tweaking for this one off did not get the same time or attention or testing.
5:04 PM
This makes sense... I think that since their usual emails are perpetually sent -- with the same content usually -- that the systems saw the one off as an imposter. The one-off also had the word "credit" within it twice along with an 800 number. Maybe these two aspects rose some flags at Yahoo!.
5:18 PM
The email filtering tech is also based on an algorithm that teaches the computer how to pick good mail. It's called Bayesian filtering and it basically looks for similarities in good emails as signs of good email, and similarity of bad emails. Like if it finds 90% of emails containing the word 'viagra' are spam, then it knows that there is a fair certainty that the new email is spam.
But if that email comes from a friend who has 347 emails not marked spam and zero marked as spam then it has to weigh that as well, would your friend send you spam when he has such a good record?
It's all in the weighting.
This is for desktop email mostly, gmail probably have different ways to handle the massive quantities of spam they process daily but I suspect the filter is similar.
So in essence you're correct, marking an email as not spam does affect the weighting somehow, but the positives are probably much less weighted than the negatives.
-- Patrick
5:36 PM
Post a Comment
<< Home