Getting Inside the Mind of an Applicant

Lara Wyatt, a high-school sophomore in Hartsville, S.C., loves to sing. She’s performed in 17 productions, playing Ursula in The Little Mermaid, Rizzo in Grease. Soft-spoken and thoughtful, she’s not the kind of 15-year-old who’s likely to worry anyone. Unless you happen to run an admissions office, in which case she might scare you a little.

For starters, Ms. Wyatt doesn’t put much stock in websites. When she began her college search last summer, she turned to Tumblr, where photos of Wagner College impressed her. She scrolled through Twitter and Instagram, seeking glimpses of campus life from “actual students,” she says, not the ones who appear in videos produced by colleges. “They seem a bit fake, especially if they are smiling the whole time.”

Ms. Wyatt searched hashtags to see where students were having fun, where they were complaining about the food. She stumbled upon a series of YouTube videos by a student at Elon University that piqued her interest in the campus. “I was able to see what college was actually like for her,” she says, “versus the videos all colleges make that say, ‘This place is awesome!’”

Although she knows that plenty of places will soon bombard her with emails and brochures, Ms. Wyatt, who expects to major in musical theater, says she’s already formed meaningful impressions, especially of Elon and Wagner, her top choices so far. She plans to continue her research on social media. “It’s so important to my generation,” she says, “to see what’s really going on.”

Today’s teenagers are grabbing the wheel of a ship that colleges have long steered. Although Ms. Wyatt, who got an especially early start, may be an outlier, her approach reveals how technology is changing the college-search process. Many prospective students don’t wait for official information, which they tend to distrust anyway. They can browse, click, and chat their way to an opinion of a campus without ever glancing at a viewbook or meeting with an admissions counselor. And they often fire off applications to colleges they haven’t previously contacted at all. In short, who will apply and who might enroll are increasingly unpredictable.

That complicates life for enrollment leaders, whose ability to meet numerous institutional goals — academic profile, tuition revenue — depends on forecasts of how many students will show up. The less colleges know about applicants, the hazier their crystal balls become. Who’s serious? Who applied only as a worst-case backup option? Such questions echo across a competitive marketplace as many administrators watch the steady decline of their yield, the percentage of accepted students who enroll.

But colleges are hardly letting go of the wheel. Instead, they’re using an array of high-tech strategies to keep control of the ship. They’re gathering more data than ever before, not just on who students are, but what they do, especially online. That includes tracking the behavior of prospective applicants as they click through a college’s website. Just as Zappos knows which sneakers to show a specific shopper, some colleges now know which major to tell a would-be applicant more about at the very moment he wants to know.

Once applications arrive, colleges are using ever more sophisticated gauges of “demonstrated interest,” a nebulous term for indications that an applicant, if accepted, is likely to enroll. Whether she’s visited the campus, liked that Facebook page, or tweeted her love for a college, some institutions track and code each move. Tallies of those actions can influence decisions about whom to admit, how much aid to offer, and where to focus scarce resources in wooing a horde of high schoolers.

Want to advance in the enrollment field? Get cozy with algorithms. Big data has barged into the modern admissions office and demanded a seat in every meeting. And as the age-old game of assessing who’s likely to enroll and succeed becomes more complex, the profession is grappling with tough questions about what strategies are appropriate — and fair.

Recruitment has reached a new frontier. Following students’ tracks online allows for personalization. Admissions offices can add new predictive variables to their projections. With all that come concerns about privacy and equity. Disadvantaged students are less likely than their affluent peers to have the abundance of time, savvy, and initiative that demonstrations of interest tend to require.

Yet the temptation to learn more about applicants is strong, and some in the field wonder just how far it will lead. Enrollment managers everywhere are looking for more of the “right” students, however defined. Knowing who really wants to come may well result in better markers of institutional standing: greater yield and retention rates. So some places are embracing the same tools corporations use to compete for customers. Sure, colleges have always been businesses, but as more data points promise to decipher applicants’ intentions, it’s fair to ask what kinds of businesses colleges are becoming.

When applicants list the colleges they’re considering, order matters. Or so some enrollment officials think. Their data-mining has convinced them that because students tend to put their top choices highest, such lists predict who’s most likely to enroll.

Dozens of institutions’ applications ask where else students are applying. Colleges also can glean that information from the Free Application for Federal Student Aid, or Fafsa. Although many enrollment officials say they use it for benign purposes — counseling applicants, research — it may factor into decisions, too. Under the National Association for College Admission Counseling’s “mandatory practices,” colleges can’t require applicants to name other colleges they’re considering or ask them to rank those institutions.

This week, at the group’s annual conference, in San Diego, voting members will consider a proposal to bar colleges from asking applicants where else they’re applying. Probing for those details is unfair to students, says Todd Rinehart, chairman of NACAC’s Admission Practices Committee and director of admission at the University of Denver. In the last few months, he says, hundreds of college counselors have emailed him to say the questions cause anxiety. Should students list colleges in order of preference, or put the place that’s asking first? If it’s important to demonstrate interest, is it OK to exaggerate it in the name of strategy? “You’re taking a 17- or 18-year-old student,” Mr. Rinehart says, “and putting them in a position where they might make a nonethical decision.”

How exactly colleges use the lists is murky; answers vary greatly. On some campuses, the information informs individual admissions and financial-aid decisions, according to NACAC officials. “There are some bad actors,” Mr. Rinehart says.

Still, those are exceptions, he and other enrollment officials insist. Most colleges, they explain, use the information not to admit or deny a given applicant, just to estimate their yield rate. “These individual predictors, taken in the aggregate, help us in shaping the class,” says Robert J. Massa, senior vice president for enrollment and institutional planning at Drew University, in New Jersey. In his experience, no single piece of information — whether an applicant visits, his or her intended major — has great predictive power. “If enrollment people are using just one thing to make a decision,” he says, “they’re misinformed.”

Several data points combined in a predictive model, however, can identify which students have a higher or lower probability of enrolling. That gives an enrollment manager a rough idea of how many students to admit and how much to discount tuition to get the right percentage of them to come. On many campuses, strong applicants’ lack of demonstrated interest isn’t going to trump their academic achievements. “Unless you have a market position where you can get away with just about anything,” Mr. Massa says, “most institutions don’t have the luxury of not admitting well-qualified students.”

Still, the perception that colleges are using students’ application lists nefariously has spread through high schools and online forums like College Confidential. Many college counselors believe colleges shouldn’t have access to such information at all. Recently, that concern prompted a change in Washington: This summer, when the Education Department released a proposed version of the Fafsa for 2016-17, colleges learned they would no longer be able to see the lists. Some institutions, the department said, were using them in ways that did not serve applicants’ best interests.

The admissions association has pushed for the change, but some enrollment officials question whether it is necessary. “It’s a gross overreaction,” says W. Kent Barnds, executive vice president for enrollment, communication, and planning at Augustana College, in Illinois. The lists are helpful to colleges, he says: not in picking students, but in planning. Among applicants who put Augustana at the top of their Fafsa list and get in, about 60 percent go on to enroll; for those who list the college second, the yield rate is about 30 percent, and third, 10 percent. “One of my jobs is to predict who’s going to show up,” says Mr. Barnds, “and this has been a really solid gauge.”

Augustana admits students on a rolling basis and does not use the information to make decisions or award aid, he says. But the lists do help determine who gets their awards first and which students the admissions staff prioritizes after acceptances go out. As students apply to more colleges, and those places receive more applications, measuring interest is one way to manage such inflation. “The same people who lament about enormously expanded applicant pools are concerned about colleges’ focus on demonstrated interest,” he says. “But I would argue that the more time and energy spent on students who’ve shown us they love us, the better it is for everyone.”

Augustana is upfront about the importance of demonstrated interest, which it describes on its website as “a key component in any student’s admissions application.” Visit the campus, get to know an admissions counselor, contact coaches or professors: All those suggestions sound like solid college-counseling advice. Other instructions, though, might sound like common sense, or read like rules for playing a game. “If we call or write or email,” the admissions web page says, “please respond.”

More colleges are tracking such responses, cataloging who reads an electronic newsletter, how many links a recipient clicks. “We’re entering a new stage of what represents demonstrated interest,” Mr. Barnds says. “The question is how to integrate it into student recruitment, how to harvest that information.”

Just after noon Central time on Thursday, September 10, a young woman in Adel, Iowa, was checking out her state’s flagship. That is, she or someone in her household had just pulled up the University of Iowa’s website.

As that person — probably the high-school sophomore herself — clicked around, a special dashboard being designed for Iowa’s enrollment office displayed her name and email address. An “engagement summary” indicated that she had visited the university’s website 49 times over several weeks, viewing 137 pages. In the bottom right corner, a “tag cloud” listed keywords associated with her browsing history (“parents,” “athletics”). The student’s overall “engagement score” — a measure of her activity on the website over the previous seven days — was a 12 (on an almost limitless scale), or “passively engaged.” Still, here were strong indications that the young woman, a couple years away from applying to college, was interested in Iowa. In the age of Amazon, indications are gold.

The system was designed by Capture Higher Ed, a company in Louisville, Ky., that specializes in “behavioral engagement.” Drawing on the same kind of algorithms online retailers use to anticipate your future wants based on past purchases, the technology helps colleges make sense of what prospective students do online, determining what information to serve them — and when. A key component: matching users’ IP addresses to information that a university already has, typically from purchases of test takers’ names and contact information. Any time someone clicks on a tagged web page, the system captures the browser’s IP address, enabling the university to track a user’s activity on the site. When a high-school student finally “opts in,” by clicking a link in an email from the university or completing an online form, the system syncs his information, and — aha! — “Anonymous User 414" becomes Joe from Sioux City.

Capture’s system helps Iowa deliver automated content, based on users’ interests, via email or pop-up messages. If an applicant has indicated an interest in business, but repeatedly views information about dance, Iowa might deliver details on both majors and its dance scholarship. A student who goes to the engineering college’s web page three days in a row might see a box noting the upcoming deadline for engineering scholarships the next time he visits. And maybe he’ll get a call from the engineering school and a student ambassador, too.

“It’s a little Big Brother-ish, potentially,” says John Laverty, senior associate director of search and prospect development at Iowa. “But all this does is say, based on what we know you’ve shown interest in, here’s the silver platter of what we think might help you.”

Iowa does not use the information in admissions or aid decisions. It’s meant to better inform applicants and their families about the university, Mr. Laverty says. In turn, the data, when combined with other information about students, can also shape his office’s recruitment strategies. “We know whether this student is really likely to enroll, or maybe, or probably not,” he says. “That helps make determinations on the amount of effort and types of communications we send.”

Besides tailoring content, Capture bills its service as a solution to a prevalent enrollment-management challenge: “stealth applicants,” who don’t engage with a college before applying. Admissions officials often say they can’t recruit students they can’t see, and those out-of-the-blue applications throw off models that predict yields.

Capture’s system is based on the idea that a college can use even an anonymous student’s online activity to craft his or her experience of a website. By providing customized content, the company’s system nudges users to share their personal information. Traditionally, a prospective student who opens an email but takes no further action might seem like a “nonresponder,” a dead-end prospect with little interest in a college; but his browsing activity may reveal that he’s highly engaged.

“Colleges have plenty of students on their sites who are interested, but they just don’t have the tools to see it, to know who’s there,” says Thom Golden, Capture’s vice president for data science. “This is an advanced listening tool. What Amazon is, we’re trying to bring to higher ed.”

Mr. Golden, who previously oversaw recruitment at Vanderbilt University, describes traditional strategies as predictable: Buy more names, send applicants more stuff, and, he says, “spam them into submission.” The future, he believes, will bring more one-on-one engagement, at least between software and student.

Moreover, understanding what an applicant does and why might become just as important as other characteristics, such as his or her geographic location and grade-point average. “Rather than conventional models that are very demographic-based,” Mr. Golden says, “we’re saying, Let’s look at students’ actual behavior.”

Some colleges have begun to plug state-of-the-art behavioral data into enrollment decisions. Although the use of such variables is limited, it’s not hard to imagine a future when they play a larger role in admissions outcomes.

Like a growing number of colleges, the University of Rochester tracks each interaction a student initiates, each response to a prompt: That includes clicking email links, calling the admissions office, participating in online chats, and even corresponding with the university via social media. Each interaction ups an applicant’s score: Once it reaches 100, a candidate is considered very serious.

When reading applications, admissions officers can see that entire history, which might include dozens of interactions spanning several months. All those data points help predict who’s a good bet to enroll, who’s probably a good fit.

Sometimes that information comes into play when an applicant’s strong but not spectacular credentials put him on the bubble. Or when a student who’s said she’s interested in music hasn’t once mentioned the university’s Eastman School of Music.

“We might look at a student’s contact history and see that every time we send them an email, they open it up and check it out,” says Jon Burdick, vice provost for enrollment initiatives. “That tells us they’re thinking about us throughout the day. That higher level of engagement matters to us.”

But it matters within a broad context. Mr. Burdick describes measures of interest as of a piece with Rochester’s supplemental essay, meant to reveal students’ understanding of the university and why they think they would be a good fit, a way to whittle 17,000 applicants down to a class of 1,300. Data science can be used for “good or evil,” he says, but there’s nothing inherently wrong with applying it to the matchmaking process.

And eagerness might take an applicant only so far. “It’s not as simple as logging onto a website each day and clicking,” he says. “None of that’s gonna work if you don’t actually get us and show us that in a concrete, personal way.”

Even if new information isn’t determinative, it may become irresistible. Many colleges invite accepted applicants to join Schools App, which links social media with predictive modeling. The app uses real-time behavioral data — how many messages a user sends, how often he or she logs in — to predict each student’s likelihood of enrolling.

Ithaca College has taken that approach a step further by inviting all applicants — and no longer just admitted students — to join IC PEERS, a website similar to Facebook, where they can connect with other prospective students and faculty members. The college tracks how many students a user friends, how many photos he or she uploads.

It’s a measure of interest that helps predict not only yield, but how a student might fare at Ithaca, says Eric Maguire, who until recently was the college’s vice president for enrollment and communication. Several years of data revealed that students who had been more engaged on the site had a higher retention rate than those who had been less active. “It wasn’t a major driver in terms of admissions decisions,” says Mr. Maguire, now vice president and dean of admission and financial aid Franklin & Marshall College. “But it was something we couldn’t ignore.”

Even as technology opens doors, it sometimes trips alarms. Consumers have more or less accepted that a company will track their online footprints to display sunglasses and cars that suit their tastes. Colleges, though, are widely perceived as different, untainted by profit margins, purer than most businesses. But the increasingly sophisticated use of predictive data and cyber monitoring in admissions reminds us that institutions, too, have bottom lines and fast-evolving ideas about customer service. For all the talk of merit and achievement, the enrollment equation also includes calculations of buyer behavior.

And more avenues for reaching students keep opening up. A recent email pitch to enrollment officials from Carnegie Communications, a Westford, Mass., company, promises to “deliver targeted online display ads directly” to screens inside prospects’ homes. It can also put virtual fences around facilities, allowing colleges to transmit messages to all mobile devices operating within a particular high school or college fair during a given time.

What kind of business is the 21st-century college? Above all, competitive and data-driven. The many algorithms of the enrollment-management universe often seem to quantify everything. One question is whether colleges will use technology to bridge divides with students, or end up subjecting applicants to an increasingly mechanized process. “One can look at it pessimistically,” says Mr. Maguire, from Franklin & Marshall, “that we’re entering into this Big Data environment and how that divorces us from the individual.”

Mr. Maguire doesn’t buy that, though. As applications proliferate, he suspects, something’s gotta give. “If I’m a recruiter, and I have several hundred kids applying from my territory, I can’t build relationships with hundreds and hundreds of students,” he says. “We need to find tools that allow us to focus our resources and build relationships with subsets of students who are most likely to be successful on this campus.”

Still, there may be limits to such predictions. Although colleges have more ways to calculate demonstrated interest, many college counselors say it’s unfair to consider, especially if students don’t know what admissions offices are counting. Putting great stock in such measurements might end up depriving a college of promising students. “It can keep you from getting the kids you really want, the kids who aren’t sweaty and desperate to get into your school,” says Mark C. Moody, co-director of college counseling at Colorado Academy, in Denver. “The kids who are actually living their lives and doing really interesting things are being punished because they’re not logging on to College Confidential constantly and finding out that they need to be replying to all these emails.”

Colleges can get carried away, too. Robert Springall, dean of admissions at Bucknell University, recalls his institution rejecting a valedictorian who hadn’t shown much love; she ended up going to Princeton. Was it right to deny a sterling applicant just to protect the institution’s yield rate? He also remembers an admissions-committee meeting where some colleagues argued about an early-decision applicant who hadn’t completed an optional essay. “You can be so focused on subtle measures of interest that you miss the big measures,” he says. It’s widely accepted that there’s no stronger measure of an applicant’s intentions than applying early.

Bucknell used to weigh demonstrated interest “aggressively,” Mr. Springall says, but it has since minimized its role in the admissions process. He worried that the university had been privileging applicants for whom it was merely convenient to visit — or who were savvy enough to know they should email an admissions officer. “Especially when talking about first-generation college students and those who didn’t have the same economic advantages,” he says, “we didn’t feel comfortable making this assessment for everybody anymore.”

Using data requires judgment. Monica C. Inzer, vice president and dean of admission and financial aid at Hamilton College, loves diving into the numbers to inform her decisions. “But sometimes the data would tell me to do things that I’m not going to do,” she says. “Every now and then, we’re gonna take a chance on a kid, and you can’t always support it, or quantify it, with data. And that’s where the holistic part of admissions comes in.”

Measures of interest — even delivered as a round number — aren’t perfect anyway. At the University of Denver, says Mr. Rinehart, some variables, like campus visits, were overpredicting outcomes. “You have to be careful with these models,” he says. “So many things can influence a score. It’s not solely a science.”

Art and science: Admissions officials often describe their work as a balance of both. In recent years, Katy Murphy thinks, it has tipped toward the latter.

Ms. Murphy, director of college counseling at Bellarmine College Preparatory, in California, has seen a diminishing emphasis on in-person recruitment as colleges rely more heavily on electronic interactions. “A large proportion of students need this personal touch,” she says. Colleges that don’t provide it, she believes, are crossing themselves off of students’ lists.

And institutions’ high-tech communications may not land well with teenagers. “They get so much stuff sent to them, email, Instagram, blah, blah, blah,” she says, “but they don’t read it, they don’t want it.”

What sways college choices in her school? Word of mouth, Ms. Murphy says. A cousin’s girlfriend who loves such-and-such campus. A great football team.

But colleges can’t leave their enrollments to chance. Robert Wyatt, president of Coker College, thinks a lot about how to use technology more effectively to connect with students. Remember Lara Wyatt, the high-school sophomore who started her college search on Tumblr? She’s his daughter. He was preparing to roll out Coker’s new website when she told him she’d ruled out some colleges based on what she’d seen on social media. “That,” Mr. Wyatt says, “was my heart-attack moment.”

The way colleges recruit students, capture their attention, and gauge their interest surely will continue to change. Enrollment offices are already looking at replacements for the long-reliable application lists on the Fafsa. One official says he plans to use some combination of the date a student applies, when she’s admitted, when she visits the campus, and maybe the number of times a family calls — all variables with predictive power. Under competitive pressure, colleges aren’t going to stop craving information on prospective students. By the law of adaptability, if they lose one tool, they’ll find another.

Eric Hoover writes about admissions trends, enrollment-management challenges, and the meaning of Animal House, among other issues. He’s on Twitter @erichoov, and his email address is eric.hoover@chronicle.com.

Correction (9/28/2015, 1:36 p.m.): This article originally referred incorrectly to a North Carolina institution. It is Elon University, not Elon College. The article has been updated to reflect this correction.