Episode Transcript
[00:00:00] Emily Chase Coleman: AI models have allowed us to take huge data sets. So you talked about
the data schools are collecting and they have so much of it and they help us to kind of sort
through that. So they help us to figure out how should we be looking at this, what are the factors
that matter? We can put a lot of data into machine learning or another AI model, but it can't be
limited to that. So there has to be a person who is thinking about the data going in and
interpreting the findings coming out. It can't just be dump the data in and the model will kind of
tell you the answers.
[00:00:42] Jeff Dillon: Welcome to the EdTech Connect podcast where we talk about all things
higher ed tech. Today we're talking about data with a guest who lives it. Emily Chase Coleman is
a higher ed data visionary who turns spreadsheets into strategy.
With 22 years in university leadership and a PhD in social psychology and statistics from
Cornell, Emily co founded HAI analytics where she now serves as CEO. Her team's AI powered
software with service platform lets colleges predict enrollment, optimize financial aid and boost
student success without a battalion of consultants.
A fearless reformer, Emily challenges the high price, high discount tuition model and champions
test optional admissions to widen access. She also amplifies women's voices in tech
entrepreneurship, drawing on her own journey building a data company from scratch. When
she's not decoding yield curves, you'll find her mentoring the next generation of women in stem.
Get ready for a fast moving conversation on predictive modeling and equitable admissions and
the future of data driven leadership in higher education.
Well, welcome to the show, Emily. It's great to have you today.
[00:01:58] Emily Chase Coleman: Thanks for having me.
[00:02:00] Jeff Dillon: I want to start by hearing what inspired you to start HAI Analytics. Like our
founders episodes are some of the most listened to episodes. So people always want to hear
how you started this.
[00:02:13] Emily Chase Coleman: Yeah, I mean, it's interesting. It's quite a journey starting a company.
I have to say that, you know, even a few years before we started hai, it never occurred to me
that I would start my own company. That seemed totally overwhelming, but the work that I was
doing and one of my colleagues we were doing together, we felt like there was sort of a solution
missing in the marketplace, in the higher ed marketplace that we were serving. And so we kind
of came up with the idea that we could author something a bit different and we just made the
leap. And that was seven years ago. And there are so many things that I think if I knew then I
maybe wouldn't have done it because you have to figure everything out. It's Just the two of us,
so we have to figure everything out. But it's been really fun. It's been a great journey. Really glad
that I did it.
[00:03:03] Jeff Dillon: So you've spent more than two decades translating raw campus data into
stories.
What was the first data set that got you hooked on analytics?
[00:03:16] Emily Chase Coleman: It's funny because I think that what got me hooked on analytics was sort
of the process of analyzing data and finding answers and finding explanations.
So the first time I sort of used a data set was in undergrad. I majored in psychology and I had to
take a research practicum. And I honestly don't even remember what my topic was. I know that
at the time it interested me, but it was more just learning how to do the statistical tests and
learning, you know, what to watch out for and how to remove bias.
That really, really kind of hooked me on analytics, and I've done it ever Since.
[00:03:59] Jeff Dillon: Yeah, your PhD work at Cornell blended social psychology and statistics.
How did that mix shape the way you now advise institutions on enrollment?
[00:04:12] Emily Chase Coleman: I think that it is. It's an important combination. And we talk a lot with our
partners about this, that, you know, there's a lot of survey work in the field. So asking students,
why did you make the decision you made? Or why did you choose the school that you chose?
And that is valuable, but can also contain a lot of biases. And a lot of times we aren't aware of all
of the factors that are influencing our behavior.
So adding the statistical piece and sort of taking the two together, where we can measure
behavior and we can look at all the factors that are quantifiable and build a statistical model to
see what's impeding a moment, and we may get an answer that's different than what if we ask a
bunch of students, why did you make your decision?
But both are very valuable. I don't think you can get the complete picture from just one. And so
it's a valuable combination, I think, that helps us understand student behavior.
[00:05:11] Jeff Dillon: Hai started as a boutique consultancy, is now a software with service
platform.
What problem on campus convinced you to productize?
[00:05:23] Emily Chase Coleman: We wanted to be able to just serve a bigger piece of the market. So not
everybody needs statistical modeling. You know, some schools have internal capabilities to do
that. Some schools are kind of just getting started in collecting and storing their data, and they
don't have the funds to kind of sign up for a contract that has a lot of services in it. So we wanted
to have an offering that was a less expensive, less time intensive for us, but that could be Used
by kind of a wide range of schools and would be helpful.
[00:06:00] Jeff Dillon: Yeah. Is it solely focused on enrollment, the enrollment side, or are there
other pockets within the university you're helping?
[00:06:07] Emily Chase Coleman: Yeah, so we focus on typically application enrollment and retention.
So we're building statistical models that kind of predict every stage of the student life cycle.
And we're looking at. We're really looking at retention as an enrollment one. So it isn't just that
schools just want to get students in the door. They want them to be successful and graduate. So
that is mainly what we do. We have done work in advancement, you know, with donor behavior.
But typically what we're looking at is enrollment and retention.
[00:06:40] Jeff Dillon: When schools compare enrollment tools, what role red flags tell you
they're buying dashboards rather than insight.
[00:06:50] Emily Chase Coleman: That's interesting. You know, I think that the important thing, even with
dashboards, is that they have to be interactive.
So you have to be able to drill down on particular populations. And if a school is saying
something like, distance from campus predicts whether students. Bone wool. So students who
are closer to home are more likely to enroll, that is a relationship that could be a correlation
without an actual pause. So you could see things like, well, actually it's because in state,
students are eligible for in state funds.
So it's really a financial issue. So kind of making these broad, sweeping generalizations without
any research behind it is a red flag. And with the dashboards, you know, we want things to be
real time. Things change really quickly, and they just really have to be interactive so that schools
can drill down on particular populations and, you know, find out really where the effect is coming.
[00:07:53] Jeff Dillon: Do you have any other stories about how you've discovered data or
schools discovered something, their data that was not expected that you mentioned? Like, yeah,
if you're closer to home, you're more likely to enroll. What are some maybe unexpected findings
you've had?
[00:08:07] Emily Chase Coleman: I think that in some of the retention work that we've done, we have found
things that are surprising factors that predict retention.
So it may be, you know, in one project that we did, we found that particular dorms had higher
retention.
And they, you know, that led to an exploration of why that was.
Was it that they were closer to classes or was it, you know, that all of the different factors. So
that was surprising at the time to that school, they didn't expect that finding. We've also looked
at course enrollment, and obviously grades are a big predictor of retention. But we have found in
some cases that taking certain courses is either a positive or a negative. And that was
Surprising. And basically what we theorized was that students who are struggling are choosing
easier classes. And, and so taking those classes is a negative predictor of retention. That wasn't
something that we expected to find. But that's what's great about statistical modeling is you can
test so many factors at once.
[00:09:14] Jeff Dillon: I want to just throw something out to you and see if how much of this you
see or that you can help with is. My take on it is that so many universities, the larger, maybe the
more exponential this problem is, but they have just these huge data sets. They just don't know
what to do with them. They don't know what questions to ask. The data is in too many silos. And
I've known other companies that have done cool stuff. And one example I've heard of was we
know in this zip code students struggle and they could track that they're having a lot of power
outages in that low income zip code and that it was affecting their, you know, in their learning
management system. They have lower scores and things like that. Questions you don't know to
ask, is that what you're doing some of is like you have all this data, you might not know what
questions to ask. Do you help these enrollment leaders see the patterns without knowing the
question to ask?
[00:10:01] Emily Chase Coleman: That's a big part of what we do. Yeah. And you know, there are a lot of
institutions that are just kind of starting their data journey or their analytical journey. And so they
started by, as you said, just reporting every piece of data that they can. And then you end up
with kind of this data overload where it's hard to figure out where to start and how to kind of
tease things apart. So a lot of what we do is helping schools figure out what questions should we
be asking?
What data do we need to answer those questions? And how do we create a system that can
both track that and we can extract it from so that we can do that analysis?
[00:10:39] Jeff Dillon: Predictive modeling can feel like a black box, I think.
How do you make the algorithms transparent for leadership like presidents and boards?
[00:10:49] Emily Chase Coleman: We, you know, one of our kind of founding principles was transparency.
And we're very open with our partners about the models that we build. We don't use kind of
proprietary methodology. We will show them the models, we will explain the factors.
Even with that, you know, there is kind of an obstacle. If someone has never built a statistical
model before or taken a statistics class, there is a piece that sometimes can be challenging to
sort of explain.
So we do our best to just sort of walk through that and just answer questions as they come. And,
you know, the more patient we are and the more patient the person asking the question, the
more likely we are to kind of, you know, meet in the middle there and be able to explain what
we're doing.
[00:11:34] Jeff Dillon: Many campuses still run a high price, high discount tuition model. What
does your data say about its shelf life?
[00:11:42] Emily Chase Coleman: I think the data say that it is nearing the end of its shelf life. We have seen
for many years that there's sort of a squeezing out of the middle class where private institutions,
they've really seen drops in the number of students from middle class families that have been
enrolling. And those are families that have too much income to qualify for financial aid according
to the formulas that have been built, but not enough to be able to afford a private education.
So they're more likely to end up at public institutions, which is fine.
You just want students to have choices with low income students. I think schools have really,
there's been a big push to serve those students and figure out, you know, how to direct financial
aid there so that those students can enroll and can thrive.
But we've gotten to this point where every family expects some financial aid they expect. If it's
not need based aid, it's a merit based aid. And parents compare, you know, what did your kid
get offered for this scholarship? And so it's gotten us into this place where the discount rates
have just gotten so high that they're not sustainable. And you know, schools have to figure out
ways to kind of lock back that discount to tell their value proposition what's the return on
investment and really make that case that the education is worth the price that they're charging
for it. So I think it's the end is near. That's my thought on it.
[00:13:16] Jeff Dillon: And now a word from our sponsor.
[00:13:21] AD: Your audiences know the true power of your institution's story.
With deep inquiry and expert strategic and creative development, Mackie Strategies applies
decades of leadership to help you drive branding, marketing and fundraising that get results.
Mackie Strategies moving your mission forward.
[00:13:43] Jeff Dillon: You know, just thinking your company's seven years old, you started pre
pandemic a couple years before and then you went through the pandemic and now here we are.
What changes have you seen in that time? Has there been this shift that's sustained versus how
these schools view data? The significance of what you're doing, Any insights there?
[00:14:06] Emily Chase Coleman: I think it has brought to light the complexity of building any sort of model
or using data that relies on past behavior to predict future outcomes. So when the pandemic
started, we could not Build a yield model based on the prior years past because the factors were
completely different.
So then schools started looking at, well, what about online behavior? You know, ping behavior,
what students are doing, how they're interacting with the school online. Maybe we can use that
because that wouldn't be affected by the pandemic, but it was in fact affected because students
are doing that more because it's their only option. So I think that that really gave everyone an
education in what you need to be careful about when you're using past behavior to predict future
behavior and how careful you need to view that. And, you know, it seems. Seems like people are
more aware of that and more aware of how they can ask those questions in a way that won't be
biased.
[00:15:06] Jeff Dillon: Yeah, well, tests optional admissions exploded post pandemic. What
equity wins and unintended gaps are you seeing in the numbers?
[00:15:16] Emily Chase Coleman: You know, I think that the wins haven't been as big as I hoped that they
would be.
So schools using, you know, test flight heavily in the admission process. We know that there's
an issue there. There's bias in the tests, and students from low income backgrounds, students of
color, do not score as highly on these tests. So, you know, it was to me a great thing when
school started falling back and really everybody sort of went test optional.
But there are other factors that are considered that, you know, are also related to income. So
even if they are basing the decision about merit aid on a kind of holistic review, what has the
student done outside of test scores? What extracurriculars do they have?
What have they accomplished that still has an economic bias? Because if you're a student who
is working to help support your family, you're probably not doing a lot of extracurriculars. So that
was a wake up call that eliminating the tests wasn't enough to get where we want to get with
that. But at least, you know, it sort of exposed that. And I think schools are working harder to
overcome those gaps.
[00:16:33] Jeff Dillon: Right, right. Do you have a success story where HAIS forecasts directly
change a financial aid award and retention curve?
[00:16:43] Emily Chase Coleman: Yes, we do. We have several. Yeah, we have, you know, a lot of clients
that we are working with. I guess all of the clients we're working with now, when they first started
working with us, the way that we determine financial aid is we have this kind of optimization
process where we're looking at, on an individual student basis, what is the aid level that
optimizes their likelihood of enrolling. And I won't get more technical than that because I know
probably no one's interested. But then we work from there to come up with the financial aid offer
and that's where we bring the retention in as well.
So what we have seen when we work with schools is that adjusting to that type of optimization
and letting that be the driver of financial aid awards often will increase yields, it will increase
enrollment, increase, you know, certain populations that they're wanting to grow. So we see, you
know, with that increased revenue and really all of the schools that we're working with now,
we've seen amount of kind of success.
[00:17:47] Jeff Dillon: You've said gut instinct, culture stalls data adoption.
What one habit should leadership teams drop this semester?
[00:17:57] Emily Chase Coleman: I think that they should drop the, you know, there is this kind of feeling or
thought process for people who have been in enrollment management for a very long time can
often feel like, I know the answers to those questions, I know what's driving student behavior,
and there's no way that numbers are going to tell me anything more about that. And I think that
is what they need to drop that. You know, we're not doing anything to replace that human
element. And actually the name of our company, HAI stands for human and artificial Intelligence.
We really believe in combining the two and the importance of that. So we're not trying to replace
anything. But we do need that part quantitative side as well to give a fuller picture of really what's
driving behavior.
[00:18:49] Jeff Dillon: How do you keep institutions from becoming dependent on your analysts?
[00:18:53] Emily Chase Coleman: That's a good question.
I don't know that we do. I mean, one of the services that we offer is that we will help schools
become self sufficient when it comes to the work that we do. We will teach someone, you know,
on their team to do it. And we have done that successfully. We have worked with schools where
over a period of a few years, they went from completely relying on us to build the models and
develop the financial aid strategy to having an internal person do it. But they have to have, you
know, they have to devote resources to that and they have to have the right person to do that.
And for a lot of schools, that is, you know, just a complication that they don't necessarily
necessarily want to deal with.
When they do decide they want to deal with it, then we can help them not be dependent on us.
We can sort of walk them through how we do what we do with the schools where we've done
that. We've stayed on in kind of a limited consultant capacity to just have that eye on the data
and know what else is happening in the field to inform the conclusions that are being drawn from
the models that they're building internally.
And schools tend to like that.
[00:20:04] Jeff Dillon: Yeah, it sounds like the philosophy of any great agency is like, they just,
schools don't have the resources right now to do everything. And I mean, more than ever,
they're really struggling with the brain drain and trying to do more with less. Right. I'm curious, on
your take with AI, how are you using AI? How's that evolved over the last few years, and what's
your take on, you know, how you see that being useful in your work with universities?
[00:20:30] Emily Chase Coleman: I think that AI models have allowed us to take huge data sets. You talked
about the data schools are collecting and they have so much of it, and they help us to kind of
sort through that. So they help us to figure out how should we be looking at this? What are the
factors that matter? We can put a lot of data into machine learning or another AI model, but it
can't be limited to that. So there has to be a person who is thinking about the data going in and
interpreting the findings coming out. It can't just be dump the data in and the model will kind of
tell you the answers. And we have seen, not necessarily with institutions that you're working
with, but with analysts that we've worked with that, you know, they're coming out of
undergraduate degrees with this incredible statistical number knowledge and this ability to build
very complex models that are very powerful. But they also have the feeling that, you know, that
sort of replaces the thinking that you have to do that. As long as you just know how to build
those models, you don't have to think about it. And that is dangerous for so many obvious
reasons.
So we make sure that that isn't happening. Like the schools that we're looking at, people, AI, you
know, it's such a buzzword and some people just have this negative reaction like it's going to
take over the world, and other people feel like it's going to solve all our problems. And I think it's
important to realize that it's not going to do either of those things, but it can help us increase the
efficiency of what we're doing if we're using it correctly.
[00:22:05] Jeff Dillon: Back to your founder's kind of journey. As a founder in ed tech, what
systematic shift would make the next Emily's path shorter?
[00:22:16] Emily Chase Coleman: Well, I think a big one would be that, you know, there's a lot of data on the
fact that female founders do not get funded at the same rate as male founders. And it's really,
really disproportionate I'm sure there are a lot of reasons why that happens, but it does put an
extra barrier in the, the way of women founders. And it's harder to kind of get in front of investors
or convince investors that, you know, this is something worth investing in.
So that's obviously a very complicated problem to solve. But, you know, if we could come up
with ways to make that more equitable, then it would make the journey much easier for, you
know, future Emily's to start the next.
[00:23:03] Jeff Dillon: All right, I have one last question for you, Emily. For presidents worried
about shrinking traditional age populations, what metric should move from quarterly to real time
on their dashboard?
[00:23:15] Emily Chase Coleman: I think my answer to that would be every metric.
I think that things change so quickly that they need to be. And maybe this isn't true at the
present level, but they need to have someone under them who's able to track the data in real
time and who is willing to shift strategy as needed. Things change very much from one year to
the next in terms of what's going to predict or what's going to drive student behavior. And if we
don't have that real time tracking of every data element we can get our hands on, then it takes
too long to make adjustments to either the admissions policy policies or the financial aid policies.
So in order to be nimble, they really have to be tracking, you know, everything that has to do
with student behavior, the application, the financial aid, that all needs to be tracked on a really
regular basis.
[00:24:09] Jeff Dillon: Yeah, I agree. Great tips for everybody.
Well, I'm going to wrap it up and that was really fun talking to you, Emily. And we will put links to
HAI analytics in the show notes as well as to your LinkedIn. So thanks for being on the show.
[00:24:23] Emily Chase Coleman: Yeah, thanks so much for having me.
[00:24:25] Jeff Dillon: Bye. Bye.
As we wrap up this episode, remember EdTech Connect is your trusted companion on your
journey to enhance education through technology.
Whether you're looking to spark student engagement, refine edtech implementation strategies,
or stay ahead of the curve in emerging technologies, EdTech Connect brings you the insights
you need. Be sure to subscribe on your favorite podcast platform so you never miss an inspiring
and informative episode. And while you're there, please leave us a review. Your feedback fuels
us to keep bringing you valuable content. For even more resources and connections, head over
to edtechconnect.com your hub for edtech reviews, trends and solutions. Until next time, thanks
for tuning in.
