The future of email engagement measurement; where do we go from here?
With the drop of iOS about a week away at the time of this writing, we’re in the final stretch of understanding exactly what the effect will be on opens. Assuming that the change plays out as the majority of the industry believes it will, opens will become unreliable for almost everyone using a modern Apple device (and thus their mail clients). The ramifications of these changes are potentially huge and affect everything from email testing, to core performance metrics, to list hygiene. In this article I want to look at the last of those, which I think is the one that is most critically affected by these pending changes.
For years now, one of the most effective best practices for list management has been removing unengaged users from your email lists. Although opening is not a true sign of meaningful engagement, it has been widely used for determining engagement for the purposes of list hygiene because it has much broader signals than clicks (meaning that you get opens more frequently than clicks). The Apple changes render opens useless as a sign of engagement, at least for the (significant) subset of recipients that will use a product protected by Mail Privacy Protection (MPP). As an aside, you might also worry that using the absence of opens to detect spam traps might suffer as well. At least for the major blocklist providers, this is not a concern – and the absence of any opens ever should still be considered a sign of a potential trap.
So, what to do and where to go from here? For most senders, the response is one of three:
- Do nothing, note that open rates will be elevated, and don’t mitigate for changes in your list hygiene approach. Ultimately, I think this approach will be ruinous for senders. Mailbox providers like Gmail don’t use your open pixels as an indication of whether a user read your mail – they have their own methods by virtue of hosting the mail, inaccessible to others. So, if your view of engagement (for the purposes of list hygiene) is completely distinct from the mailbox provider’s, you will be unable to comply with the guiding best practice of sending mail you know people want for a large subset of your mail.
- Forsake opens and move entirely to click-based engagement attribution, and consider (if you have a relationship with your recipients beyond email) looking at engagement signs outside of the email flow like site visits or app usage. For email service providers, this may be quite challenging as this is data that is typically not shared with them.
- As a pragmatic compromise, expand engagement to include both clicks and opens (as well as potentially out-of-email engagement), but drop opens that come through the MPP proxies.
As a pragmatist myself, I personally incline to option three. While it is certainly possible that the industry will increasingly move to pre-fetching images and thus render opens useless overall, in the meantime, the MPP will likely cover only 30-40% of most senders’ lists, and it seems premature to throw the baby out with the bathwater on opens. A phased approach also allows you to build models to understand how a clicks-only model compares with a clicks-and-opens model.
An interesting aspect of all of this is the real advantage that’s conveyed to senders who have deep multi-channel relationships with their users and can incorporate non-email engagement signs into determining whether a user is active or inactive (which should always be aligned to allow you to make faster decisions about when a user has gone inactive). And this is how it should be – the deeper a relationship we have with our users and the more we pay attention to the signals they send us, the better we become as communicators. And for this reason, I think MPP, while painful, will be a net positive for the industry by forcing us all to pay better attention (and give more respect) to our users.