A lot of folks are talking about Apple’s recent announcement about building privacy protection into email. I have somewhat stayed out of the conversation and I’m not sure what I really think about it. This is a change to how a lot of folks use email and no one really likes change.
I actually have a post I’ve been quietly working on talking about open rates. From my perspective, they’re an increasingly useless metric for deliverability and too many senders put too much emphasis on them and they’re not telling us what we think they’re telling us. I might work a little bit more on that, but at this point it kinda feels moot given Apple’s new announcement.
I do have some thoughts and opinions on the Apple change but I’m still thinking about it and looking at the longer term implications. I know there’s a rush to be the first with an opinion but there are a lot of angles to this. I’ve mostly been reading about what other folks have to say, so I have a bigger picture before even thinking about ‘next steps’. I know folks are upset about this and they have every right to be. But I don’t think it’s the end of deliverability or the end of email marketing. We don’t have to be happy about it, but we do have to cope with it. Just like we coped with the first spam filters and the first blocklists and the first this-is-spam button.
In no particular order here are some of my initial thoughts and impressions related to things other folks have said. Not pointing out who said what or where because, honestly, this conversation is going on in so many places I don’t even know where I saw something or who said it.
Opinion 1: Basing things like nurture campaigns and whether or not recipients get mail based on opens loses some percentage of customers. We know there are people who block remote image loading. Most webmail providers still give recipients the ability to turn it off (but I think they’ve moved away from off by default in the last few years.) Only sending mail to people who load images omits a certain type of customer / recipient from your mail stream.
Opinion 2: I’m seeing a lot of marketing statements (particularly recently) that seem to completely remove any agency from the recipient. “How do I make people open my mail” “What tricks are there to increasing engagement” “Only 30% of my recipients are opening, and how do I change the subject lines to get 35% to open?” All of these presume that something the sender does can make someone open and email or perform an action. I’ve been increasingly uncomfortable with these types of questions – it feels to me like it’s more that the recipient is a passive object to manipulate rather than a partner. A lot of what I do in delivery is treat the relationship as equal and balanced, the recipient has at least as much control as the sender.
Opinion 3: Open rates are an easy metric to measure and quantify. But that doesn’t make them a good metric. The idea that we’re now actively ‘cleaning’ open data – by trying to discover which opens are bots and which are real people, is one of the clearer signs that open rates are becoming a poorer and poorer metric. At any point where you have to remove, edit or modify data to be more accurate in your conclusions the data itself is a problem.
Opinion 3a: Apple is increasing the noise in open rates here. But we’ve known for a while that other providers (Yahoo has said they’re doing it and my personal belief is that Google may be doing this as well) are prefetching images for some emails – just like Apple says they’re going to do. The post I was working on had a lot to do with how increasingly inaccurate open rates are and that there are too many image loads that have zero to do with whether or not anyone actually opened an email.
Opinion 4: Privacy is an issue and a lot of people think it’s creepy that they are being tracked without their knowledge. Apple has positioned themselves as a company that is going to protect the privacy of their users*, even to the point of refusing to help governmental authorities to break into apple hardware to prosecute and capture terrorists. Personally, I’m waiting to see if this mail update is going to stop image loads at all. Initial information seems to indicate Apple may be removing the ‘block all remote content’ button – which is a big problem to me. I don’t want images loaded remotely for reasons that aren’t all tracking related.
(*Yes, they’re not doing this uniformly and have decided to do different things in some different countries. The world is complicated and while I disagree with their decision in general, I don’t have all the data they’re working with or know what choices they’re actually making here.)
Opinion 4a: Companies have often pushed the line of privacy invasion and tracking people. This is bigger than email – the whole big-box-store knows when you’re pregnant and will market to you before you even know you are. Said big-box-store started simply being more subtle, they didn’t stop the underlying tracking activity. They’ve also increased the amount of tracking they’re doing in store. People are, understandably, concerned that companies can pull this level of data out of their activity and that they know what you may have put in your physical cart and then removed.
Opinion 5: This is not going to be the end of deliverability and the way to reach the inbox. B2B mailing has never been reliant on engagement and open rates the way consumer mail is. In many cases business filters can’t measure engagement and in other cases engagement isn’t important for determining what mail belongs in the inbox.
Opinion 5a: In fact, the B2B space is a model for what the consumer space is starting to look like. For B2B mail we don’t have all of the data we do for consumers and yet, still, companies are successfully marketing to business domains every day. ie, we can’t fix deliverability to a business domain based on sending mail to engaged users – that’s now how the filters work. And business filters are much more prone to doing what Apple is doing – following all links in emails. We know that marketing works in this case. It’s a little more challenging, yes. But it’s not impossible.
Personally, I’ve been trying to work out other ways to capture information about subscribers for a while. And as I’ve been working more and more in the business space have been really discovering the overall limitations of ‘sending mail only to people who open messages’.
In some ways I feel like the hey day of deliverability was 2009 – 2015: probe addresses worked, panel addresses worked, open rates were easy to measure, filters were consistent across providers, FBLs were widespread, it was cool for companies to have postmaster pages, dedicated IPs and certification and the reliance on IP reputation and simple authentication all made deliverability a problem we could solve. We had a plethora of tools at our disposal and a lot of knowledge and data to base our actions and recommendations on.
Many of those tools have gradually being removed from our toolboxes in the last few years. Google and Yahoo stopped with the panel accounts. Probe accounts are not always representative of what are individual subscribers are seeing in their inbox. Postmaster pages disappeared for a lot of providers and weren’t replaced. Domain reputation got way more important and that means we have to look at increasingly complex mail through many channels to determine what is going on with delivery. We can’t just look at one channel in isolation.
I totally understand why people are upset. This is a big deal given Apple’s dominance in the email client space. Increasingly people are using the Apple mail client and they have the ability to fundamentally change email marketing by changing their mail client behavior. This is hard and scary and challenging. What do we do next? I don’t have any real insightful answers. Some of what we’re going to do is spend a couple days (or weeks for some of us) adapting to the mental shift that we’re losing such a big tool. Then we’re going to adapt.
I think, though, there’s a larger piece to think about here. The adaptation and changes coming to marketing aren’t just about email. There’s a bigger picture around online advertising and marketing and tracking and what it means for individuals. A lot of what the advertising and marketing industry has been doing is really invasive and most people wouldn’t be happy with it if they understood it. They’re starting to figure it out and governments are starting to step in and impose some rules. The power imbalance is changing and marketers are losing many of the things they’ve been taking advantage of. As a consumer, I’m OK with this. As a professional in the space I am thinking about how to adapt without violating consumer privacy.
(Copied and lightly edited from a Facebook post)
It occurred to me as I was commenting elsewhere that there is a lot of confusion about domains / subdomains and where they’re used in emails.
I get confused when people talk about ‘domain reputation’ because I have at least 4 distinct places where domains show up in an email that heavily influence delivery. But a lot of other folks talk about domain reputation as a single thing. I can never work out which place they’re talking about and thus find it difficult to comment on the domain reputation.
- SPF domain
- DKIM domain
- Visible From domain
- URL / Image hosting domains
SPF is the Envelope From / Return Path / Bounce domain / 5321.from. The end user does not see this domain unless they go look for it. It is the domain that is checked by SPF and the one that must match the Visible From domain for DMARC to pass. It does not need to be a domain controlled by the sending entity.
DKIM domain is the value in the d= of the DKIM signature. The end user does not see this domain unless they go look for it. This is intended to be a ‘domain that takes responsibility for the email’. This is the one that must match the Visible From domain for DMARC to pass. It does not need to be a domain controlled by the sending entity.
Visible From domain is what most non-email-geek people think of as the From domain. This is what is visible to the end user when they read their mail (assuming their mail client doesn’t hide it like all too many of them do). This is what consumer filters use to help drive delivery to individual user inboxes. This is the domain that is verified by DMARC.
URL / Image hosting domains are in the body of the message. This includes any links to CSS files or outside images (fonts.googleapis.com comes to mind as the big ‘shared’ domain that so much marketing mail uses).
Each of these categories develops reputation individually and then the overall email reputation is determined, in part, by how these reputations interact.
In this case I use ‘domain’ to include disparate subdomains. So I might have an email with the following ‘domains’ in it:
- SPF Domain: bounce.wttwmail.com
- DKIM Domain: tr.wttwmail.com
- Visible From: domain: wordtothewise.com
- URL / Image domains: wttw.me, image.wttwmail.com, facebook.com, click.wttwmail.com, linkedin.com, font.googleapis.com
Every single one of those domains has their own reputation and the reputation is monitored both individually and in a group.
I regularly see folks asking how to fix their Gmail delivery. This is a perennial question (see my 2019 post and the discussions from various industry experts in the comments). Since that discussion I haven’t seen as much complaining about problems.
There are steps that work to get delivery fixed at Gmail.
- Verify that your mail is actually going to bulk. I had one client that had a bad / medium reputation at Google, but their mail was actually inboxing for the most part. We spent a lot of time trying to fix the reputation without success but it didn’t matter as they were reaching the folks they needed to reach.
- Cut way back on your mail to google. Stop sending to anyone who is currently receiving the mail in their bulk folder. About the only way to know who’s getting mail in bulk is to focus on those folks who are opening or clicking on mail. Send only to people who have opened or clicked in the recent past (you can pick the timeline, but I don’t recommend going back more than 90 days for this). Do this for a minimum of a month.
- Monitor both delivery and your reputation. The reputation graphs at google are a lagging indicator and they take between 3 and 4 weeks to reflect changes in behavior.
- If you don’t see improvement: investigate what mail that you don’t know is using your domain and ensure they implement the same level of hygiene.
- If you don’t see improvement still then look at what other mail you are sending to google. There are lots of small domains that use GSuite to host their mail. Mail to those domains does affect reputation. Sometimes there’s enough volume that it breaks remediation and you need to apply the same hygiene to the hosted domains before you get an improvement in delivery.
- Once you start to see improvement in inboxing and reputation you can start to re-engage with the addresses that you removed for the reputation repair process. Do not drop them back into the feed all at once, start a warmup process to get you back up to full sends. You may need to permanently remove some unengaged recipients from the list.
It does take time to see improvements reflected in Google Postmaster tools. The good news is that when you’re on the right track mail will start to go to the inbox before you see your reputation improve.
It takes patience to fix delivery at Gmail, but it can be done. Focus on sending mail to the people you know are getting mail in their inbox and who are actively interacting with that mail. Eventually, the ML filters will learn this is wanted mail and know all of your mail should go to the inbox.
That was a longer than intended hiatus from blogging. I’ll be honest, though, talking about email just seemed so trivial in the face of what was and is continuing to happen. I posted this over on slack, earlier, and Steve pointed out I should make it public on the blog. It’s as good a way as any to come back to the blog.
With everything going on in the US, people are applying the brakes to some types of content and speech. These are not, at the moment, going to be nuanced or careful. They’re trying to stop violence, insurrection and sedition. This is potentially a place of ‘block it all and we’ll sort it out later’
I think folks should expect filters to tighten down on content – particularly political content – in the next few days and lasting for at least a few weeks. I don’t think this will be permanent and I don’t know that it’s going to affect email as much as social media and advertising. But I do think that some email systems will be affected.
There’s also a lot coming out of the martech end of things (look at what Nandini and her crew are coming out with particularly with how much fraud is in the ad networks, see https://branded.substack.com for details and links to news articles). One of the things she isn’t saying, but which is blindingly obvious to me, is a lot of the same people running the fraudulent ad networks are also in email – they’re your affiliate marketers and co-reg vendors. The fallout from the work she and her group are doing will spill into email, too.
At tomorrow’s #ltdelivery session we’ll continue talking about session: Maintaining and warming up reputations.
Invitations are going out end of the day (Dublin) today. Want to join dozens of your colleagues talking about Reputation? Sign up on our #ltdelivery page.
The next #letstalkdelivery session is Wednesday September 16 at 5pm. Invites went out today so if you signed up for our mailing list, you have the invite in your inbox. OK, if you’re on gmail it went to the promotions tab, but that’s OK, I’m promoting our call.
Check out our schedule and sign up for our mailing list so you don’t miss the next session at our #letstalkdelivery webpage.
Hope everyone had a good break.
The Let’s Talk Delivery sessions are restarting. I’ve set up a schedule and a page where you can subscribe to invites. Our next session is September 16th and we’re talking Reputation: Warmup, developing and nurturing. We talked a little last week about identity and you can follow along with the notes.
A couple things have changed about how we’re handling the sessions.
We’ve set up a LetsTalk page. There you can see upcoming questions, links to the docs and notes from current and past sessions and sign up to receive invites and notices.
Instead of collecting questions prior to the session on the Google Doc, we’ll collect them on our Facebook page. I’ll open up question pages for each session and you can add questions there.
Invites will still go out on the Monday before the session. We’ll also be opening up the Google doc for questions and feedback during the session.
These talks have been wildly successful and I am so excite to talk with all of you and discuss topics of deliverability. When I started them I thought there was some desire for peers to discuss delivery with one another. As they’ve evolved I realize they were not just open discussions but more formal training sessions. This is requiring more prep and structure now and I’m finding myself not quite keeping up with that.
Given this and the fact that it’s summer and we should all be vacationing, I’m going to be taking August off from holding the Let’s Talk calls. This will give me a chance to catch up with the infrastructure that will make this less work on my part longer term.
That means the next call will be September 2, where we’ll kick off with reputation.
You guys are awesome and I am so thrilled to be talking with all of you.
Have a great August break and we’ll talk soon.