In the Third Age, which we’re in now, ISPs factor engagement metrics into their filtering decisions, with those decisions made on the individual as well as global level. These changes mean that email subscribers not only have to tolerate marketers’ emails, but have to at least occasionally engage with them -- which in turn means that marketers can’t bloat their email lists with inactive subscribers to lower their spam complaint rates. In the Third Age of Email Marketing, the need for list quality keeps list-size ambitions in check.
However, an ISP panel at the Email Evolution Conference earlier this month seems to have muddied the water on whether email engagement affects blocking and junking. The two quasi-revelations from the panel -- which included representatives from Gmail, Outlook.com, AOL, and Comcast -- were that:
(1) Clicks don’t affect deliverability.
(2) Only spam complaints factor into blocking decisions at Outlook.com.
While not news to those in the deliverability community, these two things surprised many and caused some to question the recent emphasis that’s been placed on engagement metrics.
Don’t be confused. Here’s why you should still be concerned about engagement and inactive subscribers:
Clicks are a proxy. On the marketer side, we generally define email engagement as opens and clicks. But why would we do that when ISPs don’t use clicks to measure engagement on their end?
It’s because clicks are a proxy for all of the inbox engagement that marketers can’t see. In an interview last year, Gmail’s Sri Somanchi (who was on this EEC panel as well) revealed that the company uses “literally hundreds of signals to decide whether an email should go to the inbox or the spam folder.”
Some of the consumer signals that we know are important to ISPs include deleting an email without opening it, adding the sender to the address book, and marking an email as not junk. Marketers can’t see any of those actions. We also can’t see scrolling, read-time, forwards, or any of the other secondary signals ISPs use to make decisions.
Replies are important signals to ISPs that marketers can also see, but we’ve spent the better part of two decades telling subscriber “do not reply.” And, of course, spam complaints are an important signal that are shared via feedback loops, but those mark the end of engagement with an individual subscriber.
It’s worth mentioning that opens are hardly a perfect engagement metric either, because of image blocking. Since marketers only register opens when images render, clicks are the most common way for marketers to tell that an “unopened” email was actually read. ISPs have no such limitations and, whether images are enabled or not, have perfect visibility into if an email is read or not.
So while both opens and clicks are flawed engagement metrics, they offer marketers partial visibility into what ISPs are seeing. Drop clicks, and our already narrow field of fuzzy vision gets even narrower and fuzzier.
Junked is as bad as blocked. While Outlook.com doesn’t factor engagement metrics into its sender reputations, Gmail, Yahoo Mail, and others do. And given that trend, it’s far more likely that Outlook.com will start using engagement metrics in its filtering rather than any other ISPs will stop using engagement metrics.
But more important, during the EEC panel, Outlook.com’s John Scarrow clarified that while only spam complaints affect overall sender reputation, engagement metrics do affect individual inbox placement. So poor engagement may not get your email blocked, but it may still result in being routed to a particular subscriber’s junk folder. So, while technically different, having your emails junked has nearly the same effect as having them blocked, since most people rarely engage with emails in their junk folder.
So in the end, Outlook.com does use engagement metrics for filtering, but for individual-level rather than global filtering. The net effect is virtually the same.
The proof is the remediation. If you’ve worked at a brand that has had major deliverability problems, you probably have zero doubts about the impact of engagement metrics and the risk associated with inactive subscribers. That’s because, for several years now, purging or re-permissioning inactives has been a big part of remediation. Boosting list engagement is a proven component of getting back into the good graces of ISPs and getting blocks removed.
As painful as removing inactives is, it’s even more painful not to be able to reach active subscribers, who drive the vast, vast majority of email marketing revenue. That’s why it’s wise to have a program in place to manage your inactives that:
1. Tracks which subscribers are inactive or in danger of soon becoming inactive.
2. Tries to reengage them.
3. Reduces email volume to inactives over time.
4. Ultimately re-permissions them, removing those who don’t reconfirm their opt-in.
At the conference, Somanchi supported this framework, advising marketers to “ramp down” email frequency to inactives, and then to re-permission them in three to six months if they continue to not engage.
Better safe than sorry. When it comes to deliverability, I take a very conservative stance. I do so for a number of reasons:
1. In email marketing, it’s now way easier to ask for permission than for forgiveness.
2. Most marketers don’t have great visibility into their email deliverability, so problems can linger.
3. Blocks and blacklistings occur at the most inopportune times, such as during the holiday season.
4. Getting blocks removed takes time and money -- and that’s on top of the revenue lost from not getting your emails through.
Do you lose out on some email marketing revenue by erring on the side of list quality? Sure. But you risk more by erring on the side of list size.
ISPs are careful not to reveal too much about their filtering algorithms in order to avoid gamesmanship and keep spammers guessing. However, they have been very clear that engagement matters, whether it’s in global filtering or individual-level filtering. The onus is on marketers to heed those warnings and respond appropriately to inactivity.