Privacy Matters: Moving on from 'Opens' & how Apple shook up an industry

Andrew Bonar 0:00
To be here to speak to you. That is the very real risk the same with Kieran. So I am especially grateful for those people who came from outside of Europe because I know a lot of you have the same risk of, you know, even a false positive potentially messing up your Christmas plans. So I can't say thank you enough to Alice. She represents change.org, which, yes, there are brands, but they also just do amazing work by facilitating grassroots organizations that want to see change happen. So yeah, Alice represents a really, really nice brand is very, very close to my heart. And, yeah, I receive emails all the time, and never any spam. So Alice is someone to listen to. And she has joined with Kieran and SparkPost to help change the org send an incredible amount of emails. So looking forward to this one, and without further ado, Karen, and Alice.

Alice Cornell 1:02
Thanks, Andrew. And thanks, everyone, for staying with us. It's been a really good and fun, fun packed couple of days, I just can't tell you how good it is to be back amongst email people have really, really missed it. So thank you, everyone, for being here. We started off the conference with Karen. So I think it's only fitting that we're going to end the conference together. I wanted to invite Karen to speak with me today because I'm going to be speaking about male privacy protection, the changes that Apple have made and how that's affecting us as a sender. But how navigating that is going to be we're going to be have to work as a community to do that. And we seem to be working really closely with SparkPost. So that would be good to have your point of view up here as well. This is us, you may have guessed, I am the director of email deliverability at Change org sending quite a lot of mails. And we do that as part of SparkPost. Just sending about back do you feel everybody knows enough about this, you already don't

Kieran Cooper 2:00
know about us by now it's not my job. I'm not your technical account manager. bizarrely enough. change.org is an American organization. So looked after by one of my colleagues. But But yes, one of our flagship enterprise customers.

Alice Cornell 2:18
And I'll tell you a little bit about change to org, because you may not know much about them as you do about SparkPost. Change is the world's largest open platform, online platform for social change. And we are, we have a very clear mission, which is try and empower everybody everywhere to make the change they want to see in the world. I've worked for change for nearly nine years now. So either I'm really dull, or they just keep my keep my interest. Chase org, as I say, is the largest platform and we have over 470 million users across the world. So it's quite a lot of folks. We are staffed in 27 countries. But we actually have somebody signing petitions on our site, somebody signed petitions on a site from every country in the world, which I just I just love that there isn't a country where somebody hasn't signed a petition on change.org. And to support our users, we're sending over 600 million messages every month. So that's yes to support words, users globally. Anybody here signed a petition on change.org? Yeah, brilliant. I'd really love to hear that. In fact, in the UK at the moment, one in three people have signed on change.org users. So that's 90 million people, which is just boggling to me. Here in Spain, about 40% of Internet users use changed. org, which is fabulous. So yes, we've got some we've got quite, we've got quite some reach. But that means that we have also a lot of challenges ahead of us, especially being an open platform. Not everyone comes to the platform for the same reason. We don't have our own political agenda. So it might be that you come for there's some people who have more right wing views. Some people have more left wing views. There may be opposing petitions on the site, which is always really exciting. I'll give you a couple of examples, just about ready to talk about the Euro finals. For the Americans amongst us that soccer, football etc. But it was it didn't work out well. For the UK we lost on penalties we could have hoped it's not too triggering for you, Karen. But the bad thing that happened apart from us losing the match was that two of the lads that that what took the penalties happened to be black and they just there was horrible torrent of racist abuse that was aimed at those those folks and they're just they're just young lads, you know, playing doing their best playing a game. They're actually I mean, Marcus Rashford is an incredible activist also. Three women young women, Shyster, Huda, and Amna decided to start a petition on change.org because they just couldn't bear what was happening and they wanted people to know that not everybody in the UK is like About. And it went viral, especially in the British media. We sent one email about it to our user base. And we got 350,000 signatures just from that one email because people felt so strongly that they that this wasn't how everybody felt it was getting a lot of media attention. So it was so using Change org as the platform to amplify people's voices to say, actually, this is not okay. I'm here in Spain, we have just as much sort of media attention. In Spain. We've had lots of lots of campaigns started around the right to died in the dignified death, euthanasia. And we have lots of different petitions, starters that actually came together, to join their voices to speak to to try and get this in front of Congress, who finally after several years, a million signatures on a petition actually ratified a law to regulate euthanasia. And it was just such a powerful campaign for us working so closely with those petition starters and their families. So I just wanted to drop her on Spanish petitioning as we are here. But really what everybody wants to talk about is email. And email is absolutely critical to everything that we do. It's the lifeblood behind change.org. It's how we communicate with our users. It's how our petition starters reach their petition signers. So if they need to mobilize them to take action, maybe there's a petition delivery, maybe they're doing a Twitter campaign, we're using email to really to make that happen. And although we're exploring other channels, email, definitely not dead is you know, is still the, you know, we love email for the reasons everybody here loves email is that it's affordable, it's scalable, it's easily personalized all of the things, and we couldn't operate it without we can operate without it. So enough about change to org, you can catch me later for a drink, if you want to talk some more about it, or have questions. What we're here to talk about today is the recent changes that Apple made to their security, a calling it male privacy protection, or MPP. I'm going to give us just a quick rundown here on the top salient points for

Kieran Cooper 7:15
sure I'll try. So this is part of the latest iOS, on iPhone and iPad. And it's also part of Monterey, the latest Mac OS release. And the idea is that you can do as a user, you can opt in to do two things to prevent tracking images. But also the OS in order to make the emails load more quickly. The device is fetching images ahead of time. But more importantly, they're doing that whether or not the user is actually opening the emails, unlike Google, where for Gmail, the first time you were you open the email, images of affection then anybody who opens that email another time the images come from a cache. With Apple, they're there caching them on your device, sort of early on, and pre loading them. And that means that it's kind of having dual effects. On one hand, you don't know exactly which users have open because they're, they're not giving you the user level information. But also potentially your opens are now being increased, because the open images are being fetched, regardless of whether or not a user is opened. And as the screenshot from the the iPhone suggests it's worded in such a way that people have a generally opting in our data shows that something like 97% of iOS 15 users of opt ins opted into male MPP. For exactly the privacy reasons that I was talking about early on. It's it's an attractive thing to do.

Alice Cornell 9:05
Thanks, Kieran. No, absolutely. And we'll have a look at how this is actually affecting us at changing in the in the real world note wrote this hypothetically, when we first heard about this, once we'd stopped panicking and running around with our red hats of panic. We also spoke to SparkPost to say, Well, what do you think I think everyone was feeling their way. What does this mean? What is how is this going to affect us? Nobody really knew what was happening in great detail. So the first thing that we did as an organization was to do a full audit across the business to see what parts of the business which business units, were relying on open rate as a metric or on location data because that's also being masked. So the location we would infer from an IP, no longer necessarily available. Apple's a really big provider for us, so we knew that it was going to be impactful. Doing the audit was really interesting. We definitely found it areas of the business that we didn't know we're going to be in reacted. So for example, on changed org, you can promote a petition to share it more widely if you if you think it's really important. And the team there are using open rate for content relevancy in other ways, which we didn't even know was the thing. Our support system use the inferred IP to serve under particular legislations or in particular languages. So there were lots of different areas that I wasn't expecting this, this, this to make an impact. And so the next thing and this is also we worked closely with SparkPost on this was to try and identify the users that we're going to be impacted. How can we tell what's an apple open? What's a true Apple open? Or what is a what's a proxy open as we were calling them? And it was really useful speaking to the SparkPost team, because I think your folks would be monitoring that for a while, hadn't they?

Kieran Cooper 10:51
Yeah, so this was first announced in the summer, in a kind of press release, when the iOS was update was announced. And nobody really knew anything more than that, you know, it was just sort of Apple marketing spiel. So the only way we could do it was to download the the beta of the of the software, and then do some detailed testing to try and work out exactly what was happening. One of the one misconception early on was that this might only apply to Apple domains, you know, emails@icloud.com, or mac.com. But no, it turns out, it's to do with because it's device based, is to do with anything that anybody has set up in the mail program on their device. So you know, if like me, you have all your email accounts, coming through the mail app on the iPhone, then all of those were going to be affected whether or not they were at Gmail, Yahoo, Microsoft anywhere. So you know, suddenly, that meant that the impact was going to be a lot higher. And as the beta rolled out, and then in September, when the the full software started to be deployed, we were keeping an eye on on exactly how that was. On the the opens that we were starting to see from Apple devices, and and how many of those came from the proxy? But yes, it was, I think it was great that we had teams who were able to spend time really digging in, you know, trying to, to sort of do detailed testing to find out exactly what the impact would be. And that helped helped us admittedly, we weren't alone, lots of people in the email community were doing the same thing. And there was lots of good discussion about it. But then being able to discover that the user agent string, you know, the the thing that every, every time anything is requested on the web, there's a the user agent, that's, that's making that request, passed as a string to identify what kind of device it is. And that's used by lots of things to, you know, it's used to work out what kind of content whether to send mobile content or kind of desktop based content, it's used to identify the kind of device. And so it became clear that they were going to use a particular string, which thankfully, nobody else was using. And that was going to be one way that we could identify proxy opens. And then later, Apple released a set of IPs. So it's a list of the IPs that their proxy servers are using. So we're able to use that as another check to identify which things came from the proxy. But I'm glad the the info was useful. I think your team did the same thing we did, it was good to triangulate it,

Alice Cornell 13:35
it was great to have that confirmed, because we're sort of finding our way in there in the darkness. So once we've identified those users, we're able to exclude them from our reporting, because we knew that they would skew our open rates really, really significantly. So the only thing left to do stop monitoring the rollout. And if you're anything like me, I was obsessively looking at graphs like this every day since September, to see how these are our apple openings that are being returned, we can see that as of last week, when I think I pulled this. There was I think it was 37% of our openings being returned. Apple proxy opens which we're excluding from our open results. But of course, bearing in mind that Apple are returning and open for every mail we send into Apple's which is why it takes up such a huge share here. But if we weren't excluding them, we can see how badly that would be skewing our open rates, it would be a make it an unworkable metric. So really important to bear that in mind. It's ramped up quite rapidly. And it's and it's increasing. So I'm no idea where that's going to where that's going to end it's going to be interesting to see. And I will continue to monitor that every day. I thought this will be interesting to see that obviously Apple has different market share in different countries. These are all the countries that we deliver into and it's not listen, you know, it's not surprising. We Seeing us, Japan, UK, Australia and Canada and the top five are being impacted really significantly. Whereas other countries, Nigeria, Russia, Argentina, India, Indonesia, less so because there's there's there's less Apple uptake there. But um, so just, you know, interesting to see how that that is working out. So what were the key areas of impact that we found or that we're concerned with, I think these are probably going to be concerning to most senders. Here. The first one is that we rely heavily on email open as a as a sign of life. So we have quite strict sunsetting rules changed org, if you don't engage with emails for for three to six months, then we will just drop your email address because there's no point in sending sending email to people that don't want it. So. So we've we've been using email open as a as a sign that you know that that is a valid email address that somebody is interested that we should continue sending mail to them. And if we don't have email open, what what do we use instead? Are we just going to sunset significantly large number of of users or just much more rapidly? And even if we think about replacing that with with click rate, it's like it's it's not nearly as prolific as open rate as for us. So

Kieran Cooper 16:19
you might be sunsetting users who are actually engaging with email, but you can't tell because they're their opens come from proxies?

Alice Cornell 16:26
That's absolutely right. And so it's there's really, it's a key part of our data hygiene, we're going to have to think about new ways of how we gauge activity, how we think about an active user. And working with other data points to do that. We also use email open as a sign of content, relevance. And I know that that is a contentious point for some in the industry, you know, whether open rate is actually a sign of relevancy. I mean, you can open an email without being interested in it. In fact, you can open an email and absolutely hate it. But but we're still sort of we use that in our as a metric in our subject line testing around the world. And that's just not going to become as relevant anymore. So that's, that's a new interesting challenge for a lot of marketers, I think, and this sort of dependence on open rate, is something that's going to is going to end. And of course, we also, we've, if you've heard me or any of my team speak recently, you know, we've started on our AI journey, which we're very excited about. But we've been using open as a signal in our machine learning models, we're going to have to retrain those models to work without them. So that's, you know, it's a task but it can be done. And geolocation, as I mentioned before, we're using geolocation in lots of ways, especially to get local petitions in front of people. So you know, I might be interested in petitioners local to London. And I think that you can still get a rough idea of where a user might be based from the IP information which we get is it by country? Garin?

Kieran Cooper 17:58
I think I've heard that it's even sort of city related, obviously, dependent on on the population density. So this is the the where the proxy open happens from the IP that the proxy uses will will be roughly where the the device is located, but not as specific and obviously not an actual users IP as it used

Alice Cornell 18:22
to be. Yeah, absolutely. So lots of different areas. These are the key ones, we found others. But yeah, so what are we going to do about this? What what changes do we need to make? And the first I think there's going to need to be a real pivot towards deepening engagement with our users. And in fact, all of the things I've just mentioned previously, I think that relying on open rates has made us quite lazy as senders. And I think that's probably true for a lot of other senders, not just changed. Org, that is a sort of lazy metric is all these metrics are just proxies, but but still, an open as I mentioned, isn't really that reliable a sign, we probably want to be thinking about conversions, which in our case is people signing petitions. But we feel that this sort of pivot towards deeper engagement, understanding more about our users, being able to segment them more intelligently. So we've been using engagement to say, this is an engaged cohort, they should get more mail or this is a less engaged cohort, send them less mail. But now we're thinking about a weighted score, using lots of other data points to build a fuller picture of our users so that we can we can serve them more relevant content. And that kind of deeper engagement catching people at the beginning of their, their, their, their journey with changed org and really serving them things like that, grab their attention. And of course, if they're not engaging, then letting them go gracefully. Because not everybody wants to spend a you know, their, their time signing petitions, maybe they've just come to sign them mums petition, and that's it. And that should also be perfectly fine. We can't continue to bombard them, hoping that it's going to change their minds. But we do want to try and get relevant content in front of them early on in the journey, so that deepening of engagement. We're also doing a full review of our mail to think about how clickable is it? And when I say clickable, what I mean? What sort of value can we add to a mail that a user wants to click through to receive? I'll talk a bit about clickbait in a moment. But so we're looking to get to optimize our males for those clicks, but also the conversions. That's, that's how can we get them to take a next step? How can we provide them with with value in those emails? In fact, reviewing the males, I found some of our males don't have a call to action. Is that a useful mail? Does it help the user along their journey from signing their first petition to becoming a member of the site and sustaining us financially? So it's making us rethink what a mail journey means? And as I mentioned, exploring those alternative data points, so are they a regular user of the site? Have they promoted a petition have? Are they a member? How many petitions have they signed, we can relay that a lot of us as senators have a lot of data on our users that we're not actually utilizing. Or maybe you folks are, and it's just just us?

Kieran Cooper 21:13
No, I think lots of people, even if they have website data, they've not got necessarily got a way of joining it up with the exact email address. You know, if you have a site where you've got to log in, and the email address is sort of associated with an account, then that's one thing, although maybe email, marketers don't have access to that, that pot of data in lots of cases, but but if you don't have that kind of sign in, it's still it's then really hard to, to work out, you know, exactly which users are logging in. And to be able to use that as that, that extra bits of data, as you were saying,

Alice Cornell 21:50
That's right. It's the it's the email marketers dream is to have all of that sewn up together so that you can use everything and know everything about your customers. But easier said than done. And the lastly, thing I wanted to mention was this pivot to first party data collection. So moving away from this inferred data. And as we move more deeply into the privacy space, as Dennis was talking about, first party data collection, transparency and trust, are going to become super important. So making sure that your users, will our users trust us enough to give us their email address their phone number directly, rather than inferring it from from an IP or inferring it from from opens, etc. So I think that's going to be I think that's going to be key. Do you think I missed anything here? And

Kieran Cooper 22:35
no, I think that was excellent. Summary. Well, marvelous.

Alice Cornell 22:39
So talking about first party data collection, just at the end of September, I got this email from TripAdvisor, which I thought was really interesting. They were already out there going, right, update your details, give us your first party data and so on it, I was I, I thought that was the sort of probably a direct response to to MPP.

Kieran Cooper 23:00
But I think this one, this is exactly the sort of thing I was talking about yesterday, it's trying to demonstrate the value to the customer, of giving them giving TripAdvisor that data, you probably can't read it from here, but it's saying your email settings haven't been updated in a while in order to receive the most relevant travel deals, inspiration and recommendations, personalise your settings now, and then it's wanting even more personalized experience, click here to update your home location and see flights, restaurants and things to do specific to where you live. So it's kind of it's not just saying give us the data, it's trying to demonstrate the value to the customer in the hope that that they'll see a worthwhile transaction.

Alice Cornell 23:45
No, absolutely. That giving that yes, as we were saying giving giving that value and and being transparent about what what we're going to use the data for, and promoting that trust. So my areas of concern around this and I think concerns probably for a few of us in the industry. I think this is this could lead us to make and be practically be pressured by people like Jana Tori was talking about earlier, by people making the decisions to to adopt some probably questionable practices as a short term response to the MPP landscape. One of which being getting clicks by any means necessary in the absence of open data. So click bait having you know, subject lines that don't necessarily meet up with the with the content of the mail, etc. And as we all know, in the deliverability space, that's a really that's it's very short term can be very damaging to the sender reputation can increase complaints. And so I'm hoping that clickbait isn't going to become more prevalent as a response to this. And again, similarly, the temptation to relax data hygiene rules, already I have been People change your work do we have to sunset people who will call me just to keep them on the list for two years rather than six months, etc. And there's a reason that we do that it's good for our for our users. But it also protects our sender reputation from invalid data from spam traps from from users that just aren't that aren't real world, they are malicious people often try to add malicious signatures to change or petitions. Don't ask me why. So yes, there's temptation to take a shortcut and say, oh, we'll just we'll just keep them unless rather than now, we haven't got open opens as a metric.

Kieran Cooper 25:35
But that's a really hard case to make often, because I've had customers say to me, can you give me you know, is there any data anywhere that proves that taking that sunsetting people from your list gives you better? deliverability? And I go, No, unfortunately, but we know that it's best practice. And we know that the ISP is telling us that that is and you know, all of those sorts of rules, where people think, Well, why can't we just keep them on until they unsubscribe?

Alice Cornell 26:01
It's so true. But as any deliverability consultant knows, it's one of the first things we look at if if somebody's not reaching the inboxes, who are you sending to how how, what is the quality of your data. And I think everybody's most people have mentioned this in their presentations, today. And yesterday. What I'm interested in is that mailbox providers are going to be using open rates as a sign of positive engagement. But as senders, we're not going to be able to fully monitor that. So I'm really interested as to how that's going to have it's going to work out.

Kieran Cooper 26:36
Who knows, I mean, it's one of the things that Gmail and Yahoo, you know, people who, and presumably outlook, people who have webmail clients and or apps and are able to monitor that in their own data, and use those decisions to make filtering, use that data to make filtering decisions, as you say, we're going to be less, have less even less insight into it than we do at the moment.

Alice Cornell 27:02
So that's everything to play for. And, of course, who will be next, as as your privacy becomes more and more important. I mean, Apple have done this for a reason, as several people have mentioned already, that and Apple users are our users. So we want this set, it's the same user base, we need to be taking it seriously. And who will be next if and if another mailbox provider goes down a similar route, then it becomes even metric opens become even less of a valuable metric. So I think we really only need to prepare ourselves for a world where we don't use open metrics at all. And you've got an opportunity now to sort of bulletproof ourselves against that eventuality. It's not quite here yet, but it's, I think it's coming. And I think this is something that we need to do

Kieran Cooper 27:47
exactly. I mean, it may well be harder for Microsoft and Yahoo and Gmail, who make a lot of use of activity in those inboxes in order to generate their own products that may make them more money, which I'm sure is one of the reasons why Apple decided to differentiate themselves, because they're not in that game. But there were plenty of cynics who said, well, Apple are taking this away, they're just going to try and sell it back to us later. And that may well be true. You know, it may well be that in order to get access to information about who's opened your emails, that's going to be an extra paid service. Yeah, who are offering a lots of more paid services at the moment, you know, we've seen that in, in European mailbox providers too. So it could well be something that we have to watch out for. And so as you say, if we can get ourselves into a position of, of being able to be less reliant on it. You know, we'll we'll hopefully, it'll be less likely that those mailbox providers start to charges us for it.

Alice Cornell 28:46
Yeah, absolutely. And so if if you're, if you're a senator, and you're thinking, you know, what do I need to do? What steps do I need to take? We've certainly found it really useful to partner with with Spark Post on this, and I think most DSPs are taking steps to help their customers with this. With this.

Kieran Cooper 29:05
Yeah, we've started to based on the data science work that I mentioned earlier on, we've now put a flag into open events to indicate whether or not we think that they're prefetched. And we're going to be rolling that out into the reporting UI shortly with some kind of option to see the data with or without those opens. But we our philosophy is always to give the data to people who want to be able to to make to analyze it themselves like you guys do. So we our priority was to put that in the in the in the event data first. But I think lots of other ESPs will be doing exactly the same trying to find ways to to flag this up so that people can can view the data with without these prefetched opens.

Alice Cornell 29:55
Yeah, but I think in the long term best to plan for not having open rates or should move away from it. because as I said, it's been a discussion for some time to really sort of to really reassess what makes user engagement. What do we know about our users? And how can we, how can we best serve them? How can we give them that value, and the value of first party data and building that trust with with with our users? Finally, I just wanted to apologize, because when I was researching to do this presentation, I found this petition from four years ago on change.org, which is clearly Metatron Archangel who started it just didn't know how successful he would be. Hasn't. He hasn't declared victory on it. Yeah. But I thought it was interesting. This has been this thinking around. This has been around for a long time. Anyway, I think that's all from us. Thank you so much, Andrew, for having us. And Nely, and we'll be happy to chat about this in great detail the rest of the afternoon.

Kieran Cooper 30:57
Absolutely. Thank you.

Andrew Bonar 31:04
Thank you all so much. What a fantastic presentation to finish off the live presentations. It's Inbox Expo. It's Emailexpert. So there's always an experiment, and I'm really looking forward to this one. Our other major sponsor