Tomorrowland: The Future Of Inbox Deliverability At Top Webmail Providers

My company recently released a new study on sender reputation and the factors that influence inbox placement rates. The report showed that complaint rate, spam trap hit rate and unknown user rate are the biggest factors in determining whether or not your emails get delivered to the inbox. It's always great to have more validation for what we've been saying for years, and information like this is very powerful in helping marketers to implement best practices.

That's the world of reputation today. What does the reputation and deliverability world look like tomorrow?  As expected, I have a few thoughts.

The new big problem that ISPs and other mailbox operators are grappling with now is graymail. What's graymail? There's no universal definition, but essentially it's the mail that some people like and other people don't. At this point in the email timeline, most humans and machines recognize true spam - the messages that are pushing pills or running scams.  What is harder to discern are the kinds of emails that some people love, and others can't remember how they got on the subscriber list. Advertising driven content such as daily deals, horoscopes and newsletters may be considered graymail and are the kinds of messages that can be very hard for ISPs to handle. They may drive high complaint rates from some subscribers, but will be missed by other customers if they are blocked altogether.



Being able to block the truly bad email and recognize patterns that indicate a bad or even malicious server has become much easier for ISPs to do. Part of the reason that graymail is different and hard for ISPs to deal with is because the internal and technical structure of the email may be sound, and the reputation metrics may be good, or at least not horrible. That leaves content filtering which was never really successful on its own.

Content filtering as a first line of defense against spam just didn't work - too much real email follows the patterns of spam, and vice versa. It started an email arms race that's impossible to win, so the majority of ISPs implemented systems that made reputation the first line of defense.  Content filtering is also really expensive, especially when run on millions and millions of messages. As a result, content filtering is now applied later in the process to weed out emails that have questionable but not outright horrible reputation metrics.  This can help with the graymail problem, but it can also exacerbate it as the types of messages that are nearly evenly split between loved and loathed can be the first ones to get caught in a content filter, leading back to a false positive problem - at least for the customers who love the message.

So how are ISPs dealing with this challenge? In our conversations with top webmail providers, we see them working to solve this problem by adding other data points to create a more robust deliverability algorithm. These metrics could also allow for individual level filtering. This would allow email providers to deliver this formerly graymail to the inboxes where it is wanted, and keep it from readers that would consider it spam.

Why aren't all ISPs doing this already?  Well, for starters most of the larger ISPs are still in the very early days of figuring out how to use this finer level of sorting and detail. The danger of individual level filtering is that it could increase false positives or false negatives from a subscriber's point of view, and ISPs are reticent to make decisions on behalf of individual subscribers. It's might sound easy to assume that if someone never clicks a link in a specific company's email that they don't want the message anymore-- but that may not always be the case.

What sort of metrics could they use to drive these new algorithms and enhance their filtering techniques? Well, ISPs have access to a tremendous range of metrics and usage data. The type of information that they have access to that could provide useful include:

·       If the sender is in the address book

·       Replies to messages from a trusted source

·       Delete without opening

·       View rate (percentage of time viewing messages from a trusted source)


Marketers, of course, don't have access to this kind of information.  But that does not mean all hope is lost. You can start to look at the metrics that you do have access to in new, more sophisticated ways that will help you respond at the subscriber level.

Examining metrics by segment and thinking about your emails response rates on the subscriber level rather than the campaign level has both short and long-term benefits. In the short-term you can increase their effectiveness and help make sure your emails are targeted to specific subsets of your audience. This is going to make them perform better, right now.

But the long-term benefit is that you will be ready to respond when and if webmail providers start to implement these new metrics and begin to filter at the subscriber level. The habits you develop today will pay off when the deliverability world of tomorrow arrives.

2 comments about "Tomorrowland: The Future Of Inbox Deliverability At Top Webmail Providers".
Check to receive email when comments are posted.
  1. Bill Kaplan from FreshAddress, Inc., July 6, 2011 at 2:10 p.m.

    Great recommendations as always, George. Funny how little has changed in the past 5 years regarding the biggest factors that affect deliverability: spam complaint rates, bounce (unknown user) rates, and spam traps.

    Along those same lines, here's a MediaPost article that I wrote almost five years ago titled "Snakes in Your Database":

    Will be interesting to see what the three leading factors affecting deliverability will be five years from now.

  2. Craig Swerdloff from LeadSpend, Inc., July 12, 2011 at 4:29 p.m.

    Thanks for the great study George. I found this particular quote to be very interesting; "Unknown user rates above 1% for legitimate email is considered problematic and you should address quickly." This further supports the need for better hygiene practices including 3rd party validation services like ours at LeadSpend.

Next story loading loading..