Is Hyper-Personalization the Key to Enrollment Marketing?

[music]

0:00:12.1 Speaker 1: Hello and welcome to Office Hours with EAB. Today, we explore what AI has made possible in terms of hyper-personalized enrollment marketing. Our experts will explain what that term means and how it differs from the kinds of mail merge, direct marketing, personalization that schools have been doing for years. It’s a fascinating subject, so give these folks a listen and enjoy.

0:00:36.2 Ryan Gardner-Cook: Hello and welcome to Office Hours with EAB, the podcast where we share insights and practical guidance to help you tackle the biggest challenges in higher education. My name is Ryan Gardner-Cook, and I am new to the podcast, but I’ve been doing a lot of research lately into the topic of today’s episode, namely how institutions are “shape personalizing” the marketing messages they send to prospective students to help them stand out in what is an increasingly competitive student recruitment landscape. And that’s because students today expect highly tailored marketing messages from the schools they’re interested in. So if you were one of those institutions that hasn’t adapted to this reality, you might be falling behind the curve, so to speak. And we’re joined today by someone who is working alongside admission leaders from all types and sizes of institutions, efforts from all over the country to explore the, the limits and the best practices associated with hyper-personalized marketing. So, Michael, would you mind telling the folks who you are and what you do for a living?

0:01:56.0 Michael Koppenheffer: Thanks, Ryan. I’m Michael Koppenheffer. I’m the Vice President for Marketing innovation and AI strategy here at EABs Enroll 360 Group, which really means that I spend a good bit of my time focusing on how we use advances in technology and marketing practice to get information to students and families on behalf of colleges that is ever more engaging, relevant, specific, targeted, and so forth. So I’m excited about our discussion today.

0:02:28.3 Ryan Gardner-Cook: Great. Yeah. And let’s, let’s get into today’s topic, which has been a a major focus for both of us for some time now. And to get into it, what differentiates hyper-personalized enrollment marketing from the kind of direct marketing personalization that schools have been doing for years now, based on, you know, a few personal details they may have gathered about a recipient?

0:02:53.9 Michael Koppenheffer: Well, hyper-personalization differs from previous approaches to marketing personalization, I think in two important ways. One is a difference of degree, right? So, in the past, especially in higher education marketing, when they tried to execute it, segmentation within the marketing, they might do a version for students inside the state or students outside of the state, students who were average academic preparation, and students who were deemed high ability or so forth, and just do a couple different segments, a couple different cells, and maybe do a few of those in combination. And so the difference in how we’re thinking about hyper-personalization today is in the number of different variations and the number of different data dimensions that we bring to bear. So instead of maybe one dimension, two options, it could be three or four or five dimensions and many, many options within. So that creates much more variation. Thousands, tens of thousands, maybe hundreds of thousands of different variations for any given communication.

0:04:05.7 Michael Koppenheffer: So that’s the first difference. The second difference, which is what makes the first possible is AI, right? Is in the past we were trying to use our limited human capacities to execute this personalization, this segmentation. And there’s only so much, even the most gifted marketing strategist and the most tireless copywriter can do to bring personalization manual segmentation to light. Doesn’t mean that people don’t do it and they shouldn’t do it, but if you’re aiming for a greater degree of complexity, more dimensions, more variations, you basically need to figure out how to enlist generative AI in some way.

0:02:55.8 Ryan Gardner-Cook: Right. It’s that scalability issue. It’s not like no one has sought to take these things into consideration in the past, but especially when we’re talking at the top of the funnel here, you only have so many people on your staff to craft thoughtful messages for each student.

0:05:05.3 Michael Koppenheffer: Yes. No, you’re exactly right. It’s scalability. And I think related to that, it’s also architecture, because today the way that we think about doing personalization and segmentation in the standard old school way is by either having tokens, basically having areas within a communication where you drop in a value. So if I wanted to have an email that says, dear Ryan, I would drop in a first name token there. And that’s the sort of personalization side. And then the segmentation side is we might have areas of a communication that are dynamic. And so you designate part of the communication as static, part of it as dynamic, and you have the different variations triggered by data. Whereas in this hyper-personalized world, what we see us moving to is something that is much more synthetic. So you can basically take three dimensions, all the different variations, and actually create a cohesive single piece of communication that tees off of each of those data points in a single block. So it’s not just things that are dropped into a static hole, but it’s actually something that is more cohesive, more holistic in the way that it is personalized.

0:06:22.4 Ryan Gardner-Cook: Right. The sort of messages that a person on your school’s enrollment team might want to write themselves for each student.

0:06:32.8 Michael Koppenheffer: Exactly. Good way to think about it. Like if you had an infinite counselors writing infinite emails or notes to every student, like that’s what you’d have to do.

0:06:45.5 Ryan Gardner-Cook: So moving us along a little bit, are we mostly talking about messages incorporated into email or direct snail mail messages, or does the practice extend beyond those channels? And while we’re on the topic, I’m curious, which channels matter most to Gen Z students today?

0:07:09.6 Michael Koppenheffer: Let me take so many good things to talk about. I’ll take them in reverse order. So as we both know, AAB does some large scale survey research of high school students, college students and families every year. And one of the things we consistently ask is what communication channels matter the most for students as they are exploring colleges? And we’ve asked this question at this point for going on two decades. And the responses don’t change that much. Email, college websites are still top places to turn like physical printed snail mail is a top channel. They pay attention to paid social channels in certain degrees. And then there’s, of course, like in-person channels, if you want to call them that, like parents, counselors, friends, so forth. But email actually consistently comes up as number one. And so it is not coincidental that as we started to do our own work here at EAB on hyper-personalization, many of the pilots that we embarked on and the tests have started with email. So to be a little more specific about it, last spring, this spring just passed, we did 51 different A/B tests of hyper-personalization tactics.

0:08:36.6 Michael Koppenheffer: Most of those in one way or another involved email, either by implementing hyper-personalization within the email or using email to drive to a call to action that is in itself hyper personalized. So like a landing page that is just for Ryan that knows about your academic interest, your GPA and your location and delivers a very customized experience for you once you make the click.

0:09:06.4 Ryan Gardner-Cook: Right. So rumors of email’s demise may be greatly exaggerated given what I’ve heard out there.

0:09:15.6 Michael Koppenheffer: Yeah. Thank you for letting me wax poetic on this topic, because I am passionate about the persistence of email as an important channel, because we both have survey data. We also have engagement data that proves that students do know how to use email these days, even today’s high school students. It may not be their preferred medium for communicating with their friends, but it still plays a role in their lives and they still engage on email. They just use it perhaps differently than previous generations. And the email service providers like Gmail have made it much more complicated to try to reach new audiences than it used to be.

0:09:59.0 Ryan Gardner-Cook: Right. So if you’re still doing one of an old fashioned one size fits all drip campaign, you might be running into some deliverability issues there with that. But hyper-personalization, if I understand correctly, might help us get around that because, again, you’re sending a message that each student is inherently more interested in.

0:10:19.8 Michael Koppenheffer: Yeah, that’s what we have been seeing, which is that when you deploy a hyper personalized approach, when you’re creating lots of different variation, you’re more likely to get engagement. So that sends a good signal to the email service providers. It also means that you are not batching what Google might see as thousands or tens of thousands or 50,000 identical messages out into the universe. You really are creating something that feels more personal and tailored, not just to the student, but actually to the world of email sending as well, which will stay in good stead in terms of being a preferred communicator.

0:11:00.7 Ryan Gardner-Cook: Right, so it sounds like there are a lot of virtues to this email forward approach to testing out hyper-personalization. And in addition to that, though, email is not the only way that we are exploring and trying to push the envelope with hyper-personalization. Is that correct?

0:11:23.9 Michael Koppenheffer: No, absolutely not. So one of the other areas where we know that hyper-personalization should be able to have at least a commensurate impact for us more is physical mailings, right? A couple of years ago, our team switched our approach to how we do printing so that we are actually doing digital printing at scale, which means that we did, I don’t want to go into the nitty-gritty of printing technology, and yet I can’t help myself because that’s what enables us to be able to do hyper-personalized output in the print channel as well as in the email channel. Another channel that is absolutely ripe for hyper-personalization is texting, like SMS, because it’s in some ways even easier to execute because text messages tend to be shorter, more direct, and so you can leverage data points in a way that is even more targeted that way. So I imagine that if we were to step forward a couple years into the future, we will see colleges using hyper-personalized approaches to engage students across a variety of channels, and some that are the traditional push channels that we’re talking about and some that is a little bit more about the experience they get when they actually land on a webpage, a landing page, and so forth.

0:12:54.0 Ryan Gardner-Cook: Right, right. So baking it into every potential touchpoint between a student and a university or college.

0:13:03.5 Michael Koppenheffer: Yeah, because I think it’s going to be easier and easier to do. But the flip side of that, and this is something that I’m trying to bear in mind from years and years of hard marketing experience, is that not every segmentation actually matters. We have, at EAB, we’ve done 30 years of A-B tests of various kinds of segmentation. We’ve tried different kinds of location segmentation. We’ve tried gender-based segmentation. We’ve tried athlete, non-athlete. It is way harder than it looks to predict what someone is going to be interested in based on facts about them. It seems like it stands to reason that if you are an athlete, you are going to click on the button that’s next to the picture of the football field more, but it doesn’t always work out that way. And so what we have been trying to do as we have been rolling out hyper-personalized approaches is wherever possible, actually do them as an A/B test against a static equivalent. And the reason we’re doing that is to tell us, does it actually matter to differentiate based on this aspect or does it not?

0:14:12.6 Michael Koppenheffer: Because I think it is a false assumption to assume that everything that is personalized is therefore better and that you can predict what someone is going to like all the time based on facts about them. So we are, as I said, 51 tests in to this endeavor, but we are doing hundreds more across the rest of the year because there is so much to learn about where this matters. And it is, while it is getting easier to deploy hyper-personalization thanks to AI, thanks to the data infrastructure we’ve built, it still does not come without a cost, at least a resource cost, right? It still takes effort, it still adds complexity, and so we’re trying to figure out where it matters and where it doesn’t.

0:15:01.1 Ryan Gardner-Cook: Right. That reminds me of another one of these virtues of starting on what some people might think are a simpler way of approaching this with these emails, these email campaigns. And that’s because we already know how successful these email campaigns are. So then when we are trying to layer this hyper-personalization on top of it, we can know very quickly whether or not this is actually adding something to the message for schools and the students that they’re interested in attracting to their schools. Whereas if, say, you were working with some of these chatbots, these chatbots might be very good at automating the work that a lot of hardworking folks in enrollment offices have to do, but it may not tell us how successfully this chatbot is speaking to the student’s true intent or interest or preferences or anything like that.

0:16:03.7 Michael Koppenheffer: Yeah. I think you landed on something important, which is that it is very hard to know what is better without comparing it to something. And the more rigorously you do that, the more you can be confident that, in fact, you are improving your campaign, you’re improving your enrollment marketing endeavors, which is why we are such deep believers in the science of marketing testing and why, as we proceed with hyper personalization wherever possible, we’re doing it under a sort of testing rubric.

0:16:41.5 Ryan Gardner-Cook: Great. Now, could you explain the process that you go through in working with a school that’s just beginning to go down this road? So, in other words, how do you help them decide where to start, how deeply to jump in, and how much of their time and budget they’re going to devote to this process of hyper-personalization?

0:17:06.4 Michael Koppenheffer: Well, I can tell you a little about what we have been doing, and then I want to turn the tables a bit and ask you about some of the thinking that you’ve been doing about the ways to approach hyper-personalization over time and the different levels of sophistication that one brings to bear. Because what I would say, just to level set for all our listeners out there, is that hyper-personalization is in its infancy. Anyone who tells you otherwise is exaggerating, to say the least. The technology that makes this available is only now reliable enough and replicable enough that you actually really want to do it at scale. And the underlying data, in many cases, is not as accessible as you would want it to be to power hyper-personalization across a lot of your communications. So, as a result, when I’m talking to one of our partners about hyper-personalization, I’m talking really about two things. One is about laying the groundwork for hyper-personalization, so doing a data audit about what they do know about students at what stage, making sure that they have their data points and dimensions organized and accessible and well understood, that they know where some of the gaps are of key points of segmentation and hyper-personalization they want to execute.

0:18:29.9 Michael Koppenheffer: So, it’s really about that preparation phase. And then what we really have been doing is working with our partners on tests and pilots. Not every one of our partners is ready to tackle this right now. A lot of our partners are deeply interested and, by the same token, don’t want to be early adopters. But for the ones that are, we’ve been trying to engage them in collaborative tests of strategies and tactics that are hyper-personal that we can implement within their existing campaigns and figure out where we’re going to generate results and where we’re going to generate lift. But it is going to be a journey for all of our partners, which is why I did want to point the finger back to you, Ryan, and have you talk a little bit about the stages of hyper-personalization maturity that you developed during the course of your research and how you thought about the progression between them.

0:19:23.8 Ryan Gardner-Cook: Right, thanks, Michael. It’s the sort of thing where, as you said, this, I think, is a journey. And in understanding this as a journey where we can’t just jump to even Netflix level, obviously has their algorithms, TikTok, they have their algorithms, but they’re also dealing with a very different data environment, different levels of much more constant engagement in a lot of ways. Now, that’s not to say that we don’t have a lot of data to help us understand students here. In fact, EAB has probably the richest ecosystem of student data points out there on this subject. But it’s the sort of thing where it’s a developmental process. So, that’s why the way I’ve been thinking about this is that sort of like a staged framework or a maturity model where we start out with, if we think of the baseline as, in terms of personalization, as that manual personalization that you mentioned at the start of this, where you have an admissions counselor or someone sitting down and handwriting or typing out a message where like, I understand this student, I understand my institution, I think I can help connect with this student in a way to make the pitch for them that this is the right place for them and their interests.

0:20:51.0 Ryan Gardner-Cook: But to do this, again, especially at the top of the funnel, where I think there’s so much potential here, especially with these emails where students may not even have reached out. So, there’s no chatbot to intercede and do any work or personalization there. But at that top there, thinking about like, how can we start to scale the personalization that you had mentioned at the start to replicate to some degree what folks would be doing on an enrollment team. And so, that’s where I started to look at, well, what are the data that are necessary for that sort of personalization? And what are the data that are available to us or to our partners? And those are the sorts of things, we’re talking about like basic demographics and student preferences and things like that. Like, what major are they interested in pursuing? Where do they live in the country? Are they close by or not to this school? As we know, the distance does heavily impact whether or not a student will go to school there.

0:21:59.9 Ryan Gardner-Cook: But a couple different factors like that, which most school CRMs are collecting these data or you’re getting it from your list sources, whether that’s College Board or Apply or something like that. But most schools, if they don’t already have these data, these data are readily available to them through some sort of external partnership or something like that. And that then paired with, as you mentioned, the power of generative AI to take the raw materials of those data and also what you know about your school and the capabilities of the AI to then craft those messages for each student. We’re talking thousands and thousands and thousands of emails with hypothetically thousands of different combinations of these student characteristics. But as you said, when we are working to figure out, okay, which of these actually matters though, which of these actually moves the needle, you could be saying anything and everything about each student every time, in which case you can run into the sort of creepiness factor perhaps where it’s, well, how did they know that? Or also potentially getting it quite wrong, inferring something based on some proxy that we’re using or some signal for let’s say the student’s family income or something like that.

0:23:28.3 Ryan Gardner-Cook: But that’s the starting point, what makes this sort of personalization scalable. So that’s why we call stage one scalable personalization. But then from there, really thinking about how data availability impacts the maturity of your hyper-personalization. So at the second level, that’s where we can start working with some of those requests for information forms perhaps that a school already sends out to students in their funnel. Or especially as technology advances, perhaps you can have some smart forms on your website that collects data as you go, in which case you can take additional data that you don’t, not just the data that you get from a student’s Apply profile when they raise their hand for your school or what they reported when they took the SAT or something like that. You can start asking them more deliberate questions about, well, what are your educational obstacles? What are your real goals? What are your obstacles to attaining those goals? And you can start getting deeper into some of those interests and preferences that really drive students’ decision-making in this process. And then it’s because there are so many potential data out there, it ramps up pretty quickly from there, where when we get to our third stage of hyper-personalization, that’s where we start to get into, there’s a lot of behavioral data on students out there, even on stealth searchers, which is an increasingly large number of, large proportion of anyone’s school’s funnel.

0:25:10.8 Ryan Gardner-Cook: The kids that are not raising their hand at all, they’re not submitting any inquiry forms or anything like that, they’re just exploring the website. And that web behavior can tell us a lot about what a student’s interested in. Are they going to certain majors’ webpages? Are they going to the financial aid page repeatedly? Things like this. Did they maybe fall off and they didn’t return to your website after looking at the financial aid page? These are signals that can start to tell you, if you are ingesting these data and processing them and folding them into this communications campaign, this can tell you how and when to reach out to these students. And then when we get to stage four, and I know I’ve been going on for a while now, but this is…

0:26:00.6 Michael Koppenheffer: No, I asked for it. I asked for it. I appreciate it.

0:26:03.3 Ryan Gardner-Cook: This illustrates how much of a journey it really is, because we can’t just jump to these final stages. But that final stage would be hypothetically taking all the data you have about all these students, including these rich behavioral data, and creating truly individualized profiles about each student in your funnel, so that you know when to reach out to them on which channel, at what time, about what subject, and really help that student understand what it is about your school that could work for them and show that you as an institution are interested in this student as an individual. But it’s a long journey to get there, and a lot of it has to do with what data are available. AI has opened the doors to a lot of things, but without the right data and the right strategy on how to use those data, you might be saying something about the student, but it could come off as creepy or just not something that actually resonates with what the student is looking for without some strategic structure there.

0:27:13.2 Michael Koppenheffer: Yeah, so when you said it’s a long journey to get there, I was thinking the same exact thing. It’s like, from where we are today, and really, certainly where most institutions are today, where their approach to personalization segmentation is very manual, just getting to a point where they’re doing scalable real hyper-personalization, that is a lot of effort. That’s not going to come for every institution this year, for sure. I think that’s something that early adopters are starting to get engaged in. But beyond that, I think you, as you talked about, there’s a lot of possibility, like a lot of exciting possibility about utilizing more kinds of information, about making the messages, content, but also the structure of the campaigns really more responsive, making it more dynamic, and ultimately, like having it all be AI orchestrated. You can see this being a evolution of a marketing approach that eventually is going to take us to something that is really, really different from the intentional, structured, outbound mass approach that we have today.

0:28:29.6 Ryan Gardner-Cook: I know that you and your team have been working incredibly hard on this. And even with EAB’s relatively vast resources that we can bring to bear on this subject, is this the sort of thing that, even with the wonders of AI and the directions that it’s going, especially with these new agentic capabilities and whatnot, is this the sort of thing that you think the average school, let’s say this, the average school even is collecting more of the data that we’re talking about. How challenging do you think it’ll be for the average school to try and implement this on their own without a dedicated team?

0:29:12.2 Michael Koppenheffer: I think there are aspects to what we talked about that motivated and progressive schools are going to be trying and implementing on their own. I think there are ways in which using AI tools, using things inside of CRMs, that we will see more and more intentional personalization and segmentation happening in campaigns, for sure. It is also true based on the work that we’ve done so far that I think institutions are going to need an outside partner to help them make this happen. And I think that’s true this year. I think that’s going to be true in three to five years as well. I think the degree of technological complexity, but also the strategy complexity, is going to be required to bring this to life. It’s not going to be where admissions teams are going to want to spend their time. This feels like something that is specialized and technical and innovative to a level that I think it may become way more common, but I think it’s going to be something that colleges turn to outside partners for.

0:30:23.5 Ryan Gardner-Cook: Right. In case people didn’t already realize, this is a fairly technical subject. The idea that even if you have these data, you can just dump it into an AI and it’ll output exactly what you want. I don’t think is what the path forward looks like.

0:30:41.6 Michael Koppenheffer: I was going to say, Ryan, that I’ve been talking to enrollment teams, to higher ed leaders about AI for a couple years now. It’s interesting because at this point, many people, maybe most people, have had experiences with generative AI tools like ChatGPT. They’ve used it to draft emails for them. They’ve seen how truly nearly magical it can be in terms of independently creating pros. Sometimes people are basically like, “Well, I can get ChatGPT to write a great email to a student.” I don’t see why this is so hard to do it in the marketing campaigns, but doing it at scale across an entire campaign, across every channel in a way that is actually improving performance, that is radically different from using ChatGPT to write a note to your dentist. It really does not bear a lot of relationship. What I consistently find is that the more sophisticated the marketing and enrollment team is, the less likely they are to want to engage in that line of discussion. If they actually know how you would bring something like this to life in a hyper-personalized approach to marketing, they realize how complex it is and how challenging it will be.

0:32:00.4 Ryan Gardner-Cook: Right. And given the volume of research and commentary that, here at EAB have put out in the past year, I think it’s obvious that we’ve only scratched the surface really of what we could talk about and wax on for hours and hours. But I want to be respectful of your time and our listeners’ time here. So before we go, would you mind sharing your top advice for higher ed leaders on how to begin using hyper-personalization, the best practices, how they might measure their results and adapt and improve as they go?

0:32:38.4 Michael Koppenheffer: Big question for a closing comment. What I will say is that I think every enrollment leader should be aware that this capability is coming to the market. That if they are sticking with broad mass messages that don’t really have personalization and that aren’t using some of these hyper-personalized capabilities, they are going to be behind the curve. So I think of it as strategic orientation first and foremost. Secondly, I would find ways to get yourself educated about this topic. If you’ve listened this far in the podcast, you obviously have some commitment to do that. But beyond that, I know Ryan has written a couple white papers that help give a really helpful overview of where things are and where they’re going. We’ve both done a bunch of blog posts. I posted LinkedIn about all the time. I would find ways to stay up to date so that you are tracking where the practice is going, where the market is going. And then finally, as I’ve talked about it a couple of times, I would be paying attention to your data now. One of Ryan’s insight papers, I think it has just been released, is actually about the data side of hyper-personalization and about taking stock of where you are and where some of your gaps are.

0:34:00.7 Michael Koppenheffer: I would recommend that to everybody, just because I think it’s a no-regrets move. It’s not that hard to do ultimately, but it would give you a bit of a picture of your readiness for hyper-personalization and what you would need to do to get more ready.

0:34:15.8 Ryan Gardner-Cook: Yeah, that’s the Data Readiness Guide for Hyper-Personalized Marketing: What Student Data You Need and When You Need It. And at the end of that includes a one-page diagnostic for you to use, a self-check of sorts for either your own data ecosystem or perhaps that of an external partner that you’re considering. And that came out just this week, at least in the time of recording.

0:34:38.7 Michael Koppenheffer: Yeah, we apologize. It is a mass publication. It is not hyper-personalized to start with, but once you fill out the diagnostic, it will be hyper-personal to you.

0:34:49.0 Ryan Gardner-Cook: Well, thanks, Michael. It’s always great sitting down to talk to you about this really fascinating, I think, just growth area within the world of enrollment marketing.

0:35:02.0 Michael Koppenheffer: Thanks, Ryan. This has been really fun. As you know, I always enjoy talking about this, and it’s great to spend time together.

[music]