To Reach Students, College Marketers Prioritize AI Visibility

When Abby Isle’s daughter, now a senior in high school, began looking into colleges in late 2024, she had no dearth of information; she had toured colleges across America and Isle had bought her all the requisite guides.

But parsing that information was difficult. Isle, who has worked in technology for nearly three decades, decided to turn to ChatGPT for advice, telling the artificial intelligence tool what her daughter had liked and disliked about different institutions and the “vibe” she was looking for in her future college.

“The AI tools were able to help us say, ‘If those are your priorities … here are the schools that best fit that,’” Isle said. Most of the colleges her daughter was looking at were highly selective; ChatGPT helped direct the family toward schools with higher admit rates that still matched her criteria.

Her daughter ultimately got into Northwestern University, where she had applied early decision—in part, Isle said, at ChatGPT’s encouragement.

It was only a matter of time before large language models began serving as a college adviser. AI has become central to many young people’s search habits; a survey by the software company Adobe found that 28 percent of Gen Z users launch a search for information by prompting an AI chatbot like ChatGPT. Unsurprisingly, that includes turning to such platforms to ask questions about the college search process; a forthcoming survey from the education consulting firm EAB conducted last November shows that almost half of high school students said they were using AI in their college search process.

Optimizing for AI

The way users engage with AI is significantly different from how they use search engines. For one thing, they are likely to ask longer and more specific questions of an AI bot than they would type into a search engine.

EAB’s report found that students tend to ask AI to make lists (such as a list of nursing programs in California, for example), to help them manage deadlines and application requirements, and to help them compare different schools. Users on social media have also described asking AI to evaluate their chances of getting into specific selective colleges.

And while the end goal of a Google search is to get a user to click on a relevant link, AI platforms strive to keep the user on the site and engaging with the bot.

“There are no blue links. They’re not clicking onto websites and finding discovery elsewhere. It’s all happening with just conversations within the LLM,” said Alexa Poulin, chief digital officer at Carnegie Higher Education, a consulting firm. “That’s a massive shift, both in the information that students can gather and how they’re gathering, but also in behavior … they’re relying on AI to surface those answers versus doing their own discovery of going to multiple websites.”

Poulin said it’s imperative that colleges and universities work to ensure their information is easily and accurately pulled up by AI tools—a practice known as answer engine optimization (AEO) or generative engine optimization (GEO)—which is similar to yet distinct from search engine optimization, or SEO.

One important element of AEO is ensuring that the content available on a university’s website is up-to-date and accurate; AI can pull from old webpages that students probably wouldn’t find if they were doing their own searches, said Michael Koppenheffer, vice president of marketing, analytics and AI strategy for EAB’s Enroll360 division. Comprehensive and clear information is important, too, because AI tends to hallucinate when it needs to make inferences or fill in blanks. And institutions can also employ strategies to “tell the chatbots where to look and what information is more important,” he said.

Consultancies and educational technology firms are latching onto universities’ need for AI-optimized websites, with many now advertising AEO services. Still, it’s an imprecise science, and the same inquiry might bring up different results at different moments. Each response from generative AI is crafted in real time and therefore is influenced by a variety of factors.

“AIs are basically giant probability machines; you never know for sure what they’re going to say, and none of us can completely control what ChatGPT will say in response to a question,” Koppenheffer said. “That is both great in some perspectives, that you’re always going to get a certain amount of free agency, but also a little scary. But [there’s] nothing we can do about that. That is kind of a starting premise.”

Student Inquiries

The questions students ask of AI range widely, from broad inquiries about where they should go to college to hyperspecific questions about the financial aid, programs or campus of a specific institution.

That means colleges aren’t focused solely on ensuring the information that comes up about their institution is accurate; they also must strive for “AI visibility,” a term that refers to how likely it is that a particular brand or organization will appear in AI searches.

Chris Gage, Belmont University’s vice president for enrollment services, said Belmont hopes to reach students looking for Christian colleges in the South. But when Inside Higher Ed assumed the persona of a prospective student and asked ChatGPT to recommend such colleges, the AI didn’t recommend Belmont—even when the “student” clarified they wanted to study music business, Belmont’s top major.

“That’s a surprise,” Gage said. “If you’re the prospective student searching for music business, then hopefully … Belmont would certainly come up. It’s our largest academic program.”

However, Belmont’s marketing team has gotten better results in its own experiments with the tool, Gage noted. And when Inside Higher Ed posed the same question to Claude, another generative AI tool, it asked follow-up questions about major, institution size and Christian denomination. Inside Higher Ed selected answers that fit Belmont’s profile, and Claude suggested Belmont as one of a handful of options.

Gage noted that the university’s marketing team has identified a number of deficits in its AI visibility, such as the fact that its materials use the terms “Christian” and “Christ-centered” interchangeably. That may make it less likely to appear in a search where a student uses just one of those two terms.

Rebecca Shineman, executive director of marketing at York University, said that her institution—which is also a midsize private college—is trying to take a realistic approach to AI visibility. Institutional leaders know York can’t appear in every search, but they hope it will pop up in searches by students who are looking for the affordability and strong job outcomes that the college “is proud to excel at,” she said.

“We want to make sure from a strategy standpoint that when they’re asking these questions, we’re able to surface and ultimately that our value and story come through clearly and accurately,” she said.

When the AI was asked specific questions about Belmont, including what the “vibe” of the campus is like and what scholarships are available to music business majors, it returned responses that were accurate and comprehensive, Gage said. He was not bothered by an answer drawn from a Reddit post about how Belmont is “not a big party school,” he said.

“I think there’s always a space for a student’s authentic voice online; we want students to know the authentic Belmont, so certainly there are missional initiatives for the university that we’re going to communicate … but there’s also the lived experience for roughly 9,000 students, and that’s going to be 9,000 unique stories,” he said.

Despite the accuracy of the answers in this trial, critics note that the information coming from generative AI tools is often incorrect. Research released last fall by the European Broadcasting Union and the BBC found that AI tools made significant errors in 45 percent of responses.

‘A Little Bit of Tension’

College access leaders see both pros and cons to students using generative AI in the college search process.

Bill DeBaun, senior director of data and strategic initiatives with the National College Attainment Network, which represents college access organizations, said “there is a little bit of tension” in students using AI to stand in for college advisers, “because the college access field is deeply interpersonal. The bedrock of what NCAN members have historically done with students is asking students and families to put their trust in advisers and college access programs to provide reliable information and to help navigate a process that is difficult, foreign, uncertain for a lot of students and families.”

Keeping that interpersonal element alive in the age of AI is crucial, DeBaun said. At the same time, he can see AI being useful to college advisers and prospective students for its ability to parse large datasets quickly, meaning the tools can lead students to institutions or scholarships a given adviser may not be familiar with.

One college access organization, College Possible, is already utilizing AI in this way. The nonprofit developed its own AI tool—trained on an internal database rather than pulling from across the internet—that can answer student questions when a counselor is not available.

Shineman of York University said that she sees it as inevitable that students will use AI in the college search process, so colleges are responsible for accommodating that as best they can.

“We always talk a lot about meeting our students where they are. That can take shape in a lot of different ways,” she said. “I think today, that means including AI-powered search.”

Isle said that while she sees the benefit of college counselors, parents are best equipped to help their children with the college decision-making process because they know them best. As she sees it, AI is one way to ensure parents have enough information to then guide their children through the college search.

“I don’t have all the background on all of this, but my advantage [in] helping my kids with this stuff is I know my kid,” she said. “It helps me do the research to have a more informed opinion.”