Sunsetting policies are the strategic removal of customers from your communication stream or email program. Parting ways with customers and sunsetting communications with them can sometimes feel sad and bittersweet. After all the work you put in to acquire them, it can be hard to let them go. But it is a necessary step for healthy deliverability and a well-run email program.
We asked email experts Laura Atkins from Word to the Wise, Tejas Pitkar from Hurix Digital and Will Boyd from Simon Data to join Kickbox deliverability geeks Al Iverson & myself to discuss sunsetting customers in the age of consumer privacy protections, like Apple’s MPP. Laura kicks things off with her thoughts on Opens, an engagement metric that historically has been the key driver in determining who to sunset or when to stop sending email communications.
And off we go!
Redefining Opens & What Pixel Loads Can Tell Us
First, let’s get one thing out of the way: loading an image is not a sign that an email was opened by the user.
Yes, in the very early days of HTML email loading, the mail client loaded an image when the user opened the message. This changed pretty quickly, and users could control image loading by the early 2000s.
Conversely, for over a decade, email servers and filters have been ‘following all links in an email’ (including image links). An image load has not been associated with a user opening an email for a while now.
Referring to an image load as an open leads to misunderstanding of both the signal and what it means. I have been working hard to change my own language for clarity. Instead of using ‘opens,’ I’ve switched to the technically more accurate term ‘pixel loads.’ I encourage others to make the language shift as well.
I first started using pixel loads as a way to segment mail more than 10 years ago. I used the signal to segment mail for clients who were on the Spamhaus blocklist. When Spamhaus required addresses to be reconfirmed, I used pixel loads as a way to identify non-spam traps and people who were likely reading the mail.
I’ve continued to use pixel loads to identify addresses that were likely not spam traps and likely belonged to people who wanted the mail. They were a signal we could use as a way to minimize the pain of fixing a serious problem with a mailing list.
Using pixel loads to identify real email addresses worked pretty well overall. Clients kept a significant fraction of the folks who were reading their mail but were still able to comply with the blocklist delisting requests. It was a way to segment addresses and remove spam traps.
Many folks forget that there are, and probably always will be, users who block image loads by default. I’m one of them. All of my mailboxes are configured to never load images. Consequently, I’ve always been cognizant of the fact that subscribers will read mail without loading pixels. There are email newsletters I consistently read and enjoy but that have never seen an ‘open’ from me.
The lack of a pixel load doesn’t equal a lack of engagement with the content or a lack of desire for the mail.
The good thing about pixel loads is they are a signal and they do tell us some things about our mail. For Apple MPP and Gmail and Yahoo they tell us mail ended up in the inbox. They tell us the recipient actively uses that mailbox. A pixel load at consumer domains is a definitive sign that these subscribers received that mail in their inbox.
If only all of deliverability was that simple.
Even as we can use pixel loads to measure inboxing for consumer domains, we can’t always in a business context. In the B2B context, filters can follow images before delivering the mail to the user. But, with careful data analysis, we can use pixel loads to understand our deliverability better.
My recommendations about image load data have continually evolved. There hasn’t been any one trigger for this evolution, just an ongoing process of adaptation to changes in the email ecosystem. Pre-fetching images have been a fact for years now, Apple’s announcement just forced people to face that.
How to Build a Sunsetting Policy With MPP in Mind
Sunsetting policies have always been an interesting topic from a deliverability perspective, pre-MPP, and post-MPP.
User engagement has been crucial to the evolution of spam filtering. The more active your list, the better signals the mailbox providers have that your mail is wanted and is following best practices.
Sunsetting subscribers who don’t contribute positively to your email program helps ensure that your program gets the best deliverability possible.
This doesn’t mean you can’t mail to less active or inactive customers. It means you have to be strategic.
The predicament we find ourselves in today is that the metrics (although skewed) we once relied on so heavily to determine engagement, like opens, have changed due to the release of Apple’s MPP.
So how do you start approaching building out a sunsetting policy with MPP in mind?
- Identify what it means to be engaged
This was often as simple as, “Did they open or click?”
Was it accurate? No, but it was effective enough to build out a sunsetting plan and maintain stable deliverability.
Con: It didn’t consider those that disabled images or interacted with your email in ways that couldn’t be tracked. They also didn’t factor in the unmeasured benefits, like brand awareness, site traffic, the pile on effect of reinforced branding and exposure, and subsequent interactions with other mediums.
Although even more inaccurate, mail metrics are our closest signal on how customers interact and what the mailbox providers see, despite how they’ve changed. Opens are a good gauge that you are at least delivering to an active email address and making it to the inbox. And if you can dig into the timing of when those opens occur and weed out the obvious MPP opens, you can still get a good estimate of engagement.
Clicks are a nice compliment to the open metric as they can further paint the picture of how successful your email is. Or point out that you really need to add that CTA (call-to-action) button.
Clicks, too, can be flawed (especially with B2B email). But when you can monitor when a click occurs and if it occurs over time (users that engaged 2 or more times over the last 60 days, for example), you are able to get a clearer picture of human engagement.
Con: We are still using inaccurate metrics.
But that isn’t all we have. There are other metrics and mediums out there. Being able to identify how customers interact with your brand outside of email and how that ties back to your email program provides powerful insight that makes opens less important in decision making.
A balance is needed between what the mailbox providers see and what you see is driven by your email program. Finding this balance is the tricky part, but you can start with a few questions to identify other signals that can show you how your email program is performing:
- Does your site traffic increase after a send?
- Does search increase after a campaign?
- Are abandoned carts revisited or do purchases improve?
- Customize communications for customers in the twilight of your relationship
For me, Pre- and Post-MPP can follow these guidelines.
As your customers start to fade, but are still expressing some interest, you may want to reduce cadence, try different messaging, or a re-engagement campaign.
When to do this really depends on your business model. I’ve seen some businesses do this after 30 days, 90 days, 9 months, and beyond. To determine your best timeframe, test, test, test. Then monitor and measure what happens:
- Do your campaign performance metrics drop/increase?
- What about revenue, conversions, new signups, or your other KPIs?
- Can you attribute a drop in performance to the lack of communications? Or is it steady without change? If it’s the latter, you may have started to find out when your customers are fading.
Besides cadence, customize your content. Run tests, ask questions, and ask for feedback. What is it that will make your content relevant and timely again?
If you are able to re-energize your customers, don’t just put them back in your active file. They lost interest for a reason so flesh out the why and address that.
- Were there too many touches? If so, bring them back at a reduced cadence.
- Are they a seasonal customer? Reduce cadence during off-season times and ramp up as you get closer to re-engagement.
- Is the content lacking relevance? Incorporate personalization and preferences from form data, website activity, or preference pages.
- Sunset those truly unengaged
Throughout the interwebs, you’ll find advice stating you should begin your twilight and sunset initiatives after a year of no engagement. This works well from a campaign performance perspective. But also because of list churn and a general timeline for when emails are eligible to be transitioned into spam traps.
Con: It didn’t consider other data points or customer lifecycle or those that suppress image loads.
Some may stick to the same ‘year’ cutoff. And although you likely heard some experts pushing against this pre-MPP, you’re likely hearing this more. Data has become a big word post MPP release, and words like first-party and zero-party data.
So long as you aren’t seeing a degradation in performance, your complaints are managed, and your reputation and deliverability are strong, your sunsetting timeframe may look much different.
Despite MPP, I still think sunsetting is a wise decision. Again, carefully consider all of your customer signals, but also understand that keeping a list of 5-year inactive subscribers without any signals will land you in hot water on the deliverability front.
And if you find this audience provides an ROI benefit you can’t part with, separate that stream to another subdomain and IP and use email verification to help weed out churned emails. But beware, your success on that IP may not be high.
Outside of deliverability, there are some legal considerations for how long you should retain a person’s PII. GDPR has requirements that data should only be retained for as long as is strictly necessary. There are some exceptions, but retention for marketing purposes is not one.
Once you identify those that have sailed off into the sunset with no signs of turning around, it’s time to bid them adieu. If interest comes back, make sure they can sign up again.
I wouldn’t be creepy and suddenly mail them as soon as they come across your site, but use that data to customize the experience and communications while on your site or connecting through other mediums or if they decide to opt-in to talk again.
Before you begin sunsetting, make sure your database and systems are flexible enough to re-enable customers should they choose to come back. An unsubscribe attribute flag set apart from a user-driven unsubscribe can do this.
You may want to try one last effort to win them back, but don’t belabor the farewell. In some cases, this final parting is a great way to remind them they haven’t interacted in a while. And if you can fit it in without being awkward, remind them why you became bosom buddies in the first place.
But when do you finally pull the plug?
MPP inflating opens may delay some accounts being sunset, but we know that if there is no interaction, those are great candidates for removal. The ‘when’ depends on your business.
Generic hard and fast rules can easily constrain businesses with customer relationships that have long life cycles—car or technology purchases. And if decisions are made in a vacuum focusing on email metrics alone, you likely will be abandoning customers before their time has come.
And, to add a slight contraction to the recommendation, sometimes, there are scenarios where a business really has no insight into activity. This could be because they send emails with few CTAs and they are sending to B2B domains or another provider that proxies and disrupts tracking–none of which provide complaint data to remove uninterested recipients either.
In these cases, sunsetting may not be in your best interest (in the traditional sense). Instead, focus on regaining permission, collecting replies, and offering easy ways for your customer to let you know what they want (easily accessible unsubscribe options/preference pages/etc.).
If you aren’t sure, TEST.
TLDR; There comes a point where saying goodbye outweighs the benefit of having a large, but sluggish list. But the best sunsetting policy for you is the one that works for your customer relationship.
You Don’t Need a Sunset Policy: You Need Better Audience Mangement
Sunset policy defines identifying your unengaged subscribers and deciding if it’s okay to send them emails or not.
Sunset policies are defined by email marketers to maintain a productive and engaged list along with safeguarding deliverability.
Sending repeated emails to cold subscribers brings down your opens, clicks, and conversions affecting your revenue.
Mailbox providers see such behavior as potential spamming, downgrading your sender reputation and inbox placement.
My take on sunset policy?: You don’t need a sunset policy but a better audience management strategy.
Here’s how to tackle low-interest subscribers and get them back:
- Segment your list according to tiers of audience engagement. The top tier will be the hyper-engaged customers who regularly open, click, and explore.
- The next tier should be moderate engagers who respond only with certain mailers and need more tailored content.
- The last tier should be for unengaged users who need reminding from time to time of your brand and need careful nurturing.
Here’s how to engage with the last tier:
- Reduce your frequency to the dormant users. If you were mailing them once a week, then shift to once a month.
- B2C senders, take advantage of sale seasons’ spike in activity to engage with your subscribers. For B2B, this could be events, webinars, and podcasts you are hosting.
- Target them in batches and segment them into smaller chunks. Sending to cold users in smaller volumes helps your deliverability and reputation.
- You can also use different shared IPs and domains to send this type of communication. This would avoid damaging your primary domain used to send marketing communication. Make sure the engaged are getting added back to your regular list.
- Offer them exclusive discounts and content to get them involved. A special message or offer will help them.
- If the inactive segment shows low interest, engage them with other channels – ABM, Social media, Linkedin/FB ads are always available and provide results. Email is not the only solution.
- For those last-tier audiences, you can segregate them into further tiers according to non-responders. Set them aside to send them communication, say once a quarter or 6 months when some big event/sale will be around the corner.
- If still unengaged, send them a quasi-break-up email where they can unsubscribe or downgrade their preferences. Either way, you are giving them a chance to ‘opt-out’ rather than ‘report to spam’.
- Keep updating the list by verifying email addresses using a service like Kickbox.
Yes, Apple MPP might have affected opens, but they don’t affect your customer intent. Switch your monitoring metrics to clicks and conversions, so you don’t get sucked into the whirlpool of proxy/bot opens.
Trusting opens is not an option but if you wish to count on an open, look at other signals the users are sending to form a better opinion.
Sunsetting Subscribers in the New Normal
Apple’s Mail Privacy Protection (MPP) surely did change open tracking quite a bit. Can you trust your open tracking nowadays?
Anywhere from 15-50% (or more!) of your tracked opens are coming through Apple’s proxy process now. And many, if not most, of those are not “true” opens—they’re just Apple’s proxy bot tickling the open pixel, without any accompanying user-initiated action.
Here’s how I visualize what’s happening here. I think of an email campaign as a glass, a water glass or a tumbler. Its full-height represents all of the subscribers you sent that campaign to.
Last year, before MPP, after sending, the glass would fill up with water, to about 40% full. That 40% full, representing what percentage of subscribers opened your email message.
Still slightly inaccurate due to proxy and bot activity, but good enough for government work. Note that this leaves about 60% air in your glass – about 60% of your subscribers who are not engaging. No click, no open.
Now, after MPP, when you send, the glass would end up more like 75% full of water. But that water is not just opens. It’s now composed of opens + MPP proxy opens from Apple users, and it is now vastly more inaccurate. You can’t easily break down that 75% to identify which are which. It’s a mess.
But that 25% of air – non-openers? It is mostly accurate. In that campaign, 25% of people definitely did not interact. Now, some percentage of Apple users, represented in the 75% water (opens), did not actually interact.
So your 25% doesn’t capture EVERY subscriber who failed to interact with your email message. But it caught a lot of them.
Meaning, it’s still safe to trust the “lack of engagement” signal based on the lack of click or open after a certain period of time.
In the past, it would have reliably caught MORE unengaged subscribers than it does today. But it will still catch a good chunk of them. If you suppress mail to them, you’ll help improve your sending reputation by boosting your engagement rates.
Of course, strategy still applies here. Don’t just sunset at random. Define a timeframe and a process.
How long before you define a subscriber as inactive? And then what do you do about it? It’s going to vary by industry – what are you selling or sending? In many cases, it might make sense to consider somebody unengaged if they haven’t opened or clicked in 90 days.
Maybe after 90 days, you drop them into a series of win-back campaigns, looking to offer up a compelling call to action in exchange for a click (a sign of life) from the subscriber.
Maybe you just keep unengaged subscribers in a separate bucket and only email them in small amounts, every month or every few months.
The best strategy is really going to vary, but the single most important thing to keep in mind is that an email address doesn’t last forever.
People abandon email accounts, and people abandon senders. If they’re ignoring you and you keep sending to them forever, it will eventually drag down your engagement rates enough to cause deliverability issues.
So do keep in mind that you do need to do SOMETHING here, and that strategy and sunsetting is still possible, even with Apple MPP in the picture.
Evolving Sunsetting Policies for the Modern Inbox
As I’ve worked with senders over the years, my advice on sunsetting has evolved. I’ve always been adamant that senders who do not pay attention to how recipients are engaging with their messages are destined for a life of deliverability problems.
For some time, it seemed like the only senders eager to take this advice were ones who had experienced some sort of deliverability pain like a blocklisting.
However, in recent years it has become more and more common for email marketers to utilize a sunsetting policy as a proactive part of their email deliverability toolkit rather than just a reactive repair strategy.
This growth in sunsetting adoption seems to stem from both an increased focus on email engagement by mailbox providers and spam filters and the increased availability of segmentation tools for marketers, which make implementing these policies much easier.
Tools for segmenting an audience have come a long way in recent years. It is much more common than ever before for senders to be able to target recipients based on a wide variety of data points.
This gives them great flexibility in providing personalized content and journeys designed to move a recipient through a lifecycle rather than simply building a list of email addresses that are blasted with messaging into perpetuity.
With this increased ease of sophistication, I find myself working with senders to move away from having one simple sunset policy that applies to all sending toward sunsetting logic applied differently across different mailbox providers, mail types, address collection source, recipient journeys, and more.
Just like one message rarely fits all in terms of your audience, one simple sunset policy across a sender’s entire marketing program is no longer sufficient.
The more intentional a sender can be with how they actively manage the relationship with their recipients versus simply trying to avoid deliverability problems, the more effective and sophisticated they can become with their sunsetting policies.
While segmentation tools have made crafting sophisticated sunset policies easier, the industry has seen other advancements making things a bit more murky for email marketers.
Advancements like Apple’s Mail Privacy Protection make the email open event a less trustworthy indicator of human engagement than ever before.
With opens increasingly not a reliable way to determine if a recipient is truly engaged with messages, it is important for marketers to find ways to incorporate other, first-party data points into their sunsetting logic to help more accurately identify unengaged recipients.
Senders who can attribute data, such as site visits or app usage, LTV change over time, purchase frequency changes, or any other data that indicates waning brand engagement to email campaigns, can use this data to supplement email deliverability stats to boost the efficacy of sunsetting efforts.
While it is still the case that having some “line in the sand” logic that prevents sending to very unengaged addresses is a helpful tool, the more a marketer can use tools available to craft recipient journeys and experiences informed by more than email deliverability data, the more effective marketers will be at crafting email programs that reach the inbox and the recipient.