AI Mode now has over one billion monthly users. Queries are doubling every quarter. The new Intelligent Search Box takes video, images, files and your open Chrome tabs as inputs. It runs on Gemini 3.5 Flash. It can spin up “information agents” that crawl the web on a standing brief and report back. It can build a working mini-app on the fly. Gemini Spark, a personal AI agent that lives in the cloud, has its own Gmail address, and will potter about doing tasks for you while you sleep.

This is the biggest change to how people find information since Google was a verb. And almost none of the commentary so far has been about what it means for higher education.

Why this Google I/O is different

This isn’t another platform panic. It’s a structural change to the part of the funnel that decides applications.

Every few years our industry has a collective panic about a new platform. Snapchat. Stories. Reels. Threads. BeReal. Each time, the same conference circuit, the same agency decks, the same “we need to be where Gen Z lives.” Each time, three quarters of it turns out to be a rounding error. This is not one of those times, and the reason is structural.

Search is the bottom of every applicant funnel ever drawn. A sixth-former in Doncaster does not start by following Loughborough on TikTok. They start by typing “best universities for sport science UK” into Google. Then something else. Then something else. By the time they fill in a UCAS form they have run somewhere between forty and two hundred queries. That is the part of the funnel where they decide.

Google has just rebuilt that part of the funnel from the studs up. It is going to take a while for the rest of the world to catch up to what that means, because the news doesn’t have a logo or a creator economy attached to it. But it is the most important thing that’s happened to HE marketing since Clearing went digital.

THE HE TAKEAWAY The channel that decides your applications has just changed. If your strategy doesn’t have a view on this in writing by the end of June, you’re not in the conversation.

What actually happened

Three changes from yesterday’s announcements actually matter for HE.

First, the search results page is being replaced by a conversational interface that summarises blogs, news sites, social posts, real-time data, and your own Gmail into a single recommendation. You will still get blue links if you scroll. Don’t count on people scrolling, you’ll loose that bet.

THE HE TAKEAWAY A serious chunk of your prospective applicants will never see your website with human eyes. Plan accordingly.

The philosophy shift we’re not talking about

The question is no longer how do we get found. It’s how do we get recommended, and the answer is governance, ethics and values, not channels.

Here is the bit I think is going to matter most, and the bit hardly anyone is writing about yet.

For thirty years, the strategic question in digital marketing has been: how do we get found? Pages. Keywords. Backlinks. Bidding. Audiences. Lookalikes. Whole careers have been built on the mechanics of visibility.

That question is about to be replaced by a much harder one: how do we get recommended?

Being found used to be a competition between pages. Being recommended is a competition between reputations. And reputations are weird things. They can’t be bought in an auction. They don’t move on a Monday morning because somebody changed a meta description. They accumulate over years, across thousands of small signals, and they’re the one thing an AI cannot manufacture for you.

That distinction is going to matter even more because of what AI does to the rest of the marketing stack. When anyone with a laptop can produce a polished prospectus video, a beautifully written course page, a convincing student testimonial and a synthetic open-day photoshoot in an afternoon, the value of any one of those assets in isolation collapses. The signal-to-noise ratio of the internet is about to get a lot worse before it gets better. Universities are going to be competing for attention in a world where nobody, not the applicant, not the parent, not the AI agent doing the shortlisting, can trust what they’re looking at on face value.

In that world, the things that have always been the soft, slow, hard-to-measure parts of brand stop being soft, they actually become operational.

How do you govern yourselves?
What are your ethics?
What do you actually value?
Whose side are you on when it matters?
What do you say no to?

These are not abstract questions for a brand book any more. They are the questions an AI agent is implicitly answering on your behalf every time it summarises you to a prospective student. The institutions that have clear, defensible, public answers to those questions are about to find that they accidentally built the most valuable marketing asset of the next decade. The ones that don’t are about to find out what it feels like to be a generic prospectus in a world of infinite generic prospectuses.

This is why I think the right response to AI Search isn’t to chase the technical mechanics of GEO, important as they are. It is to use this moment to get the brand right. Stand for something specific. Stand for something other institutions don’t. Get trust banked now, while it’s still relatively cheap to earn, because in eighteen months it is going to be the only currency that matters.

THE HE TAKEAWAY The work you’ve been putting off, values, governance, point of view, is now your most important marketing investment. Trust is the only asset AI can’t generate.

Why “it’s just the new SEO” is wrong

Old SEO competed on pages. AI Search competes on recommendations, and recommendations are built from the whole brand surface.

There is a comforting line going round. It goes: “AI Search Visibility is just the new SEO.” Treat it like SEO and we’ll be fine. Schema markup, structured data, FAQ pages, job done.

This is wrong in the specific way that most reassuring things are wrong. It is technically true and strategically useless.

SEO worked because Google rewarded the page that best answered the query. AI Search works because Google synthesises an answer from across the entire web, then decides what to recommend. The unit of competition is no longer the page. It is the recommendation. And the inputs to the recommendation are everything an applicant has ever heard about you, everywhere on the internet, weighted by how trustworthy each source looks.

If a prospective student asks AI Mode “should I go to Sheffield Hallam or Leeds Beckett for marketing”, you do not win that with better SEO. You win it with more evidence, across more places, that a serious person would choose you. Reviews. News coverage. Mentions in industry reports. Graduate outcomes data. Specific course content that survives summarisation. Public answers to the questions in the last section. The whole brand surface.

The sector’s real problem

Most HE marketing teams already know all of this. The gap is between knowing and doing.

Here is the bit I think gets us all stuck.

Most of the HE marketing directors I talk to understand this stuff intellectually. They’ve read the reports. They’ve seen the AI Mode demos. They know the funnel is changing. What they don’t have is the internal cover to do something about it.

Because doing something about it means saying difficult things to senior leadership. It means telling the academic faculty that their course page is going to get summarised by a machine, and the machine isn’t impressed by the dean’s preferred adjectives. It means rerouting budget away from the agencies who’ve been pitching the same prospectus refresh for five years. It means picking a point of view in public, and that means somebody, somewhere in the institution, will object.

The default response to that level of difficulty is the working group. The strategy refresh. The internal communications exercise. Eighteen months later, a slide deck exists, three people have moved on, and the website still doesn’t have transcripts on the videos.

The sector is, frankly, suffering together. Everyone knows what needs to happen. Almost nobody has built the internal momentum to make it happen. And the institutions that break out of this pattern in the next twelve months are going to look unrecognisable to the ones that don’t, by 2028.

THE HE TAKEAWAY Your problem is not knowing what to do. Your problem is making it happen inside an institution built to discuss things, not ship them.

What we think universities should do

Five practical moves you can start this quarter.

I have spent most of my career being suspicious of agency people who write provocative things and then offer no advice. So here is the advice.

1. Stop briefing creative as if humans are the only audience. Half your audience is now an LLM, and an LLM cannot watch a 90-second hero film. It can read a transcript. Put one on every video. Put structured data on every course page. Write the bits humans will skim as if they need to survive being summarised by a machine, because they do.
2. Audit your reputation surface, not just your website. Go and look at what Reddit, The Student Room, Which? University, the Guardian league table, your local press and your alumni LinkedIn say about you. Then ask yourself: if Gemini synthesised all of this, what would it recommend? If the answer is “not us”, that is a brand problem, not a channel problem.
3. Get specific about what you stand for, in public. Not on a values page nobody reads. In the course content, the comms, the press releases, the staff posts. Pick a few things you genuinely believe and put your name to them. The institutions that try to be everything to everyone are about to disappear into the AI summary as “another good university with strong links to industry.” The ones with a point of view will get cited.
4. Controversially, be careful about just spending more on paid media right now while you work out which channels still convert when half the journey runs through AI Mode. The temptation in moments of uncertainty is to spend faster. Resist it for a quarter or two. The data we’re getting back on AI Mode behaviour in the next six months is going to be more valuable than another £200k against a creative concept built for the old funnel.
5. Make sure someone owns this internally. Reading the announcements, talking to applicants, briefing the website team. Most agencies, ours included, do their best work when the client has someone in-house who can match us thought-for-thought. This is the year to give that person the brief.

The uncomfortable part

The craft economy of HE marketing is about to be valued at near zero by the people doing the shortlisting.

A lot of higher education marketing for the last decade has been about producing content. Beautiful prospectuses, polished video, atmospheric photography, carefully managed social grids. It has been a craft economy. It has supported a lot of agencies, including this one, and a lot of in-house teams.

An agent like Spark, asked to recommend a university, is going to weigh that craft economy at zero. It will weigh course content, outcomes data, third-party reviews, student-generated evidence and your public reputation at one hundred.

Start making the unglamorous, structured, evidence-rich stuff an agent can use. And treat brand strategy as the operational discipline it has just become, not the bit you do once every five years.

If that sounds dull, I’m sorry. The future of HE marketing is going to be about a third brand, half evidence, and the rest technical infrastructure nobody puts in a showreel. We are going to have to rebuild bits of our offer, and so are most of our clients. The teams that get there first are going to have a quietly enormous advantage.

What we’re doing about this at Education Cubed

We started building toward this picture about eighteen months ago, because the writing was on the wall. Three things have come out of that work that we’re now offering directly to universities.

We’re deliberately not selling these as one big transformation programme. Most universities don’t need that. They need momentum on one specific thing, then the next, then the next.

If you read this far and any of it lands, the cheapest way to start is a 45-minute call with us. No deck, no pitch. We’ll look at how your top three courses currently appear in AI Mode and tell you what we see. If there’s nothing there for us to help with, we’ll tell you that too.

Three questions for Monday morning

If you only do one thing after reading this, put this meeting in the diary.

Forty-five minutes. Your head of marketing, your web lead, your admissions lead and one person from recruitment.

• When did we last look at how our courses appear in AI Mode? If the answer is “never”, you have your starter.
• What is our point of view on agentic applicant journeys? Not next decade. The next eighteen months.
• What do we actually stand for that another university doesn’t? If you can’t answer that one in a sentence, the AI can’t either.

The teams that have this conversation in May are going to be in a very different place by September. The ones that wait until results day to ask will be playing catch-up for the next two years.

If you need an AI brain in the room, that’s exactly what I’m here for.

Email me at wgreeves@educationcubed.com to find out more and book your free 45‑minute AI Mode review.