When Search Stops Sending Traffic: What Zero-Click Means for Higher Education

For most of the history of digital marketing, we operated under a fairly simple assumption: people search for something, they get a list of results, and if we have done our jobs reasonably well, they click on one of those results and arrive at our website.

Higher education has spent an enormous amount of time and money working within that model. We redesign websites, obsess over program pages, debate URL structures, research keywords, improve page speed, rewrite titles, add schema and periodically undertake SEO projects that produce spreadsheets large enough to require their own governance committees.

All of that made sense because the click was the bridge between discovery and engagement.

Increasingly, the bridge is disappearing.

A SparkToro and Datos study found that 58.5% of Google searches in the United States and 59.7% in the European Union ended without a click. In other words, nearly six out of ten searches produced no visit to any result at all. More recent research suggests the number has continued to rise, although changes in the underlying data make direct comparisons imperfect.

Now, zero-click search itself is not particularly new. Google has been answering weather questions, displaying sports scores, showing definitions, surfacing maps and putting featured snippets above search results for years. If I want to know how tall the Eiffel Tower is, there is very little reason for me to visit the International Association of Eiffel Tower Measurement Professionals to find out.

What is different now is the type of question that can be answered without a click.

Generative AI can synthesize information, make comparisons, apply criteria and produce recommendations. Google is putting AI-generated answers directly into search results, while ChatGPT, Gemini, Claude, Perplexity and others allow someone to skip the traditional search-results page entirely.

That distinction matters quite a bit to higher education.

College search is particularly well suited to AI

Think about how prospective students actually research colleges.

Very few are searching only for "University X tuition" or "University Y application deadline." Those are easy questions, and search engines have been capable of answering them for a long time.

The questions that matter during discovery are much messier.

A student might be looking for a medium-sized university in the Northeast with a strong nursing program, decent financial aid, access to a city, a campus that does not feel enormous and an admissions profile reasonably close to their own. A working adult might want an online master's program that can be completed part time, does not require the GRE, has a credible reputation and will actually improve their prospects of moving into a different career.

Historically, answering those questions required a lot of research. Search. Open five tabs. Read a few program pages. Go back. Search again. Find Reddit. Open another six tabs. Forget which school said what. Start a spreadsheet. Abandon the spreadsheet.

AI is very good at removing that friction.

And students are figuring this out quickly.

In a 2026 EAB survey of more than 5,000 high school students, 46% said they were already using AI during the college search process. Among those using it, 62% had used AI to identify colleges that might be a good fit, and roughly half had used it to research application requirements or discover institutions they had not previously considered. Most interestingly, 34% said AI research had increased their interest in an institution, while 18% said something they learned through AI caused them to remove a college from consideration.

Parents are moving in the same direction. An EAB study released in August found that 57% of parents surveyed had used AI to evaluate and compare colleges, and about a third said AI had helped them discover institutions they had not previously considered.

So this is no longer really a theoretical discussion about what prospective students may do someday.

They are doing it.

The important change is not search. It is who constructs the answer.

This is the part I think higher education needs to pay particular attention to.

For years, institutions have assumed that if they could get a prospective student to their website, they could tell their story.

And they could.

The institution controlled the program description. It decided which outcomes to highlight, which photographs to use, how to explain the student experience, how prominently to feature accreditation, faculty expertise, career outcomes, financial aid, research opportunities and everything else it considered relevant.

AI changes that relationship.

If a prospective student asks ChatGPT to compare five MBA programs, the student is not necessarily reading five MBA program pages. ChatGPT is reading—or has learned from—information about those programs and constructing the comparison for them.

The institution is no longer necessarily telling the story.

The AI is.

And the answer may be assembled from the university website, government data, rankings sites, news coverage, Reddit conversations, third-party directories and any number of sources the institution neither owns nor controls. Sometimes the university's own website may not even be among the sources used. We have written before about why strong Google rankings do not guarantee an institution appears in AI answers, and about how often AI systems describe institutions inaccurately or omit them entirely.

That presents a very different problem from traditional SEO.

Being number one in Google meant you had a very good chance of getting the opportunity to explain yourself.

Being represented in an AI answer means someone else has already done the explaining.

Your website becomes both a destination and a source

None of this means university websites are suddenly irrelevant. Quite the opposite.

Students still place considerable trust in institutional websites. Research from UPCEA and Search Influence found that 77% of prospective students surveyed considered university websites the most reliable source when confirming information they found elsewhere.

But the website is beginning to serve two audiences simultaneously.

The first is the audience we have always designed it for: people.

The second is the collection of systems those people are increasingly asking to interpret the institution for them.

That has some practical consequences.

A beautifully written paragraph about an institution providing "transformative educational experiences that empower tomorrow's leaders to achieve their fullest potential" may have survived countless rounds of committee review, but it tells an AI system almost nothing useful.

What programs do you offer? Who are they for? What do they cost? How long do they take? What are the admissions requirements? Are they online, in person or hybrid? What makes one program meaningfully different from similar programs elsewhere? What careers do graduates enter? What evidence supports the claims being made?

Those are much more useful pieces of information, for prospective students and for the systems trying to answer their questions.

This does not mean writing robotic content for robots. In fact, I think that would be precisely the wrong lesson to take from all of this. It means being unusually clear about facts while still being interesting when explaining them. It is also worth understanding why a site that looks well optimized for LLMs may still not be reaching them.

Higher education websites have spent years accumulating layers of marketing language, organizational language and content written largely because a particular department needed to have a page. AI search creates a pretty good incentive to finally ask whether that content actually helps someone understand the institution.

Probably not the worst exercise for humans either.

We also need to rethink what we measure

This may be the harder adjustment.

If a prospective student asks an AI assistant for the best programs matching a particular set of criteria, sees your institution recommended, learns several things about it and adds it to a list of schools to investigate, that interaction has value.

Your analytics platform may know absolutely nothing about it.

There may be no session. No pageview. No referral source. No click at all.

Eventually that student might visit the website directly, search for the institution by name, respond to an email or appear in the CRM as an inquiry, but the moment of discovery happened somewhere else and will probably receive no credit.

For twenty years, organic traffic has been one of the easiest proxies we had for digital visibility. It is becoming a less complete one.

That does not mean we should stop measuring organic traffic, any more than the arrival of Google meant we stopped caring about direct traffic. It means we need additional measures for a new channel: whether the institution appears in relevant AI answers, how frequently it appears, which sources are being cited, whether the information presented is accurate, what attributes AI systems associate with the institution, and which competitors appear when the institution does not.

This is sometimes called Generative Engine Optimization, or GEO. As is customary in technology, we apparently cannot experience a meaningful change without immediately manufacturing an acronym for it.

But the underlying idea is important regardless of what we eventually decide to call it.

Search visibility used to mean being present in the list of places a prospective student could go for an answer.

Increasingly, visibility means being present in the answer itself.

This does not mean SEO is dead

I am generally suspicious whenever someone announces the death of a technology or discipline that is still being used by several billion people.

Students still use Google. They still visit university websites. They still watch videos, browse social media, read rankings, talk to friends, attend college fairs, visit campuses and—occasionally, I am told—even read the emails enrollment offices send them.

AI is not replacing the college search process. It is inserting itself into it.

That distinction matters because the right response is not to abandon SEO and chase some entirely new collection of GEO tricks. It is to recognize that content now travels further than the page on which it was published.

A useful program page can generate traffic. It can also supply facts to an AI Overview, support an answer in ChatGPT, become a citation in Perplexity, reinforce information found on another website and influence a comparison the institution never sees.

The page may be doing more work while appearing to do less of it.

And I suspect that is where higher education marketing teams are going to have to become a little more comfortable with ambiguity. Not every useful interaction will produce a click. Not every influence will produce an attributable session. Some of the content that matters most may increasingly do its work somewhere other than the institution's own website.

The question is therefore changing.

For years we asked: Can prospective students find our page?

Now we also have to ask: When they ask someone else about us, what answer do they get?

That is a much harder question.

It is also rapidly becoming one we cannot afford not to answer.

About TondroAI Discover

TondroAI Discover was built to help higher education institutions answer exactly that question. It audits how major AI systems describe an institution, where the information is accurate or wrong, what sources are shaping those answers, and where the institution is absent from conversations in which it reasonably should appear. The goal is not to promise control over AI answers—nobody can honestly promise that—but to give institutions visibility into a discovery channel that conventional web analytics cannot see.

If you would like to see what AI systems currently say about your institution, get in touch.