Did the algorithm change?

When faced with unexplained deliverability changes one of the first questions many folks ask is “Did the algorithm change.” In many ways this is an meaningless question. Why? Because there are two obvious answers to the question.
A1: Of course it didn’t.
A2: Of course it did.
Both answers are correct, but they’re answering different underlying questions. When we understand how two diametrically opposed answers are both correct, we understand much more about filtering.

What is an algorithm?

When we’re talking about spam filtering the algorithm is the process or rules to follow.

Basically, an algorithm is a computer program that is set up to filter spam to the bulk folder and filter wanted mail into the inbox.  This algorithm doesn’t change. It can’t.
In the case of many (most?) spam filters, the filters incorporate features of machine learning. 
This means the algorithm is constantly changing, learning more and more about what is spam and what is wanted mail based on user interactions.

Algorithms don’t change

Overall, the algorithms don’t change that frequently. They are fed data (lots of data) on a continual basis. They take feedback from recipients (spam / not spam buttons) and developers (new data sets of known bad and known good mail) to learn what good mail looks like and what bad mail looks like. But the underlying code doesn’t change very frequently.

Results change

Machine learning algorithms are only as good as the data they’re fed. In the case of spam filters, the input data is constantly changing. So the output results change. Sometimes an email that was not-spam one day is spam the next because the algorithm caught up with a new threat or new behavior. 

Delivery is still in your control

This machine learning and reliance on end users to help tune filters may make it seem like spam filtering is completely out of the senders’ control. That there is nothing a sender can do to get into or out of the bulk folder. The good news is, the underlying algorithms are pretty simple: wanted mail goes to the inbox, unwanted mail goes to the bulk folder. As with everything, details matter. Senders who are focused on recipients usually don’t have a difficult time reaching the inbox. Companies that focus on themselves and try gimmicks find it much harder to consistently reach the inbox.
 

Related Posts

What do you think about these hot button issues?

bullhornIt’s been one of those weeks where blogging is a challenge. Not because I don’t have much to say, but because I don’t have much constructive to say. Rants can be entertaining, even to write. But they’re not very helpful in terms of what do we need to change and how do we move forward.
A few different things I read or saw brought out the rants this week. Some of these are issues I don’t have answers to, and some of them are issues where I just disagree with folks, but have nothing more useful to say than, “You’re wrong.” I don’t even always have an answer to why they’re wrong, they’re just wrong.
I thought today I’d bring up the issues that made me so ranty and list the two different points of views about them and see what readers think about them. (Those of you who follow me on Facebook probably know which ones my positions are, but I’m going to try and be neutral about my specific positions.)

Read More

From the archives: Taking Permission

From February 2010, Taking Permission.

Permission is always a hot topic in email marketing. Permission is key! the experts tell us. Get permission to send email! the ISPs tell us.
Marketers have responded by setting up processes to “get” permission from recipients before adding them to mailing lists. They point to their privacy polices and signup forms and say “Look! the recipient gave us permission.”
In many cases, though, the permission isn’t given to the sender, permission is taken from the recipient.
Yes, permission is being TAKEN by the sender. At the point of address collection many senders set the default to be the recipient gets mail. These processes take any notion of giving permission out of the equation. The recipient doesn’t have to give permission, permission is assumed.
This isn’t real permission. No process that requires the user to take action to stop themselves from being opted in is real permission. A default state of yes takes the actual opt-in step away from the recipient.
Permission just isn’t about saying “well, we told the user if they gave us an email address we’d send them mail and they gave us an email address anyway.” Permission is about giving the recipients a choice in what they want to receive. All too often senders take permission from recipients instead of asking for permission to be given.
Since that post was originally written, some things have changed.
CASL has come into effect. CASL prevents marketers from taking permission as egregiously as what prompted this post. Under CASL, pre-checked opt-in boxes do not count as explicit permission. The law does have a category of implicit permission, which consists of an active consumer / vendor relationship. This implicit permission is limited in scope and senders have to stop mailing 2 years after the last activity.
The other change is in Gmail filters. Whatever they’re doing these days seems to really pick out mail that doesn’t have great permission. Business models that would work a few years ago are now struggling to get to the inbox at Gmail. Many of these are non-relationship emails – one off confirmations, tickets, receipts. There isn’t much of a relationship between the sender and the recipient, so the filters are biased against the mail.
Permission is still key, but these days I’m not sure even informed permission is enough.

Read More

Thinking about deliverability

I was chatting with folks over on one of the email slack channels today. The discussion was about an ESP not wanting to implement a particular change as it would hurt deliverability. It led me down a path of thinking about how we think of deliverability and how that informs how we approach email.
The biggest problem I see is the black and white thinking.
There’s an underlying belief in the deliverability, receiving, and filtering communities  that the only way to affect sending behavior is to block (or threaten to block) mail.

This was true back in the ancient times (the late 90’s). We didn’t have sophisticated tools and fast CPUs. There weren’t a lot of ways to handle bad mail other than to block. Now the landscape is different. We have many more tools and the computing capacity to quickly sort large streams of data.
At most places these days, blocking is an escalation, not a warning shot. Many places rate limit and bulk folder questionable mail as a first strike against problem mail. Sometimes the mail is bad enough to result in a block. Other times, it’s not bad enough to block, so it disappears into the bulk folder.
There’s a corresponding belief in the sending community that if their behavior doesn’t result in blocking then they’re acting acceptably. This isn’t true either. There are a lot of things you can do (or not do) that don’t help delivery, but will actively harm delivery. Likewise, there are things you can do that don’t actively harm delivery, but will help. All of these things add up to reaching the inbox.

Read More