Higher education is investing in AI in student recruitment. The hard part is knowing where it works.
The pressure on university recruitment teams has never been greater. Student expectations have shifted dramatically, shaped not just by the rise of consumer AI tools, but by a broader expectation that institutions communicate the way other trusted brands do: fast, personal, and across whatever channel the student happens to be on that day.
The response from many institutions has been understandable: invest in AI. The problem is where that investment is landing.
Across the sector, money is flowing into the most visible applications – student-facing chatbots, workflow tools labelled as AI, and features that signal technological progress to senior leadership. These are easy to procure, demonstrate, and understand. They’re also, in many cases, the wrong place to start.
The infrastructure problem that we all need to be talking about
The reason AI deployments in student recruitment underperform isn’t the technology, it’s what’s underneath it.
Over years of evolution, student recruitment infrastructure has accumulated CRM systems, admissions platforms, communication tools, and data repositories that rarely talk to each other. AI depends on reliably structured data, consistent workflows, and connected information to function effectively. When those foundations aren’t there, even the most sophisticated tools produce noise rather than insight.
This isn’t a niche technical concern, it’s the central challenge. The question isn’t just “Which tool should we buy?” It’s “Do we have the data infrastructure to make any tool worth buying?”
What a student journey actually looks like
One student we supported to enrolment last year, a 25-year-old from Mauritius enrolling in an MSc in Digital Marketing and Analytics in the UK, interacted with the university 159 times over nine months. Those interactions spanned 74 nurture emails, 32 outbound one-to-one emails, 25 inbound emails, 22 WhatsApp conversations, and six phone calls. She enquired about entry requirements, chased scholarship results via WhatsApp, attended a visa information session, and had a final call with an adviser before enrolling.
The question AI needs to answer in student recruitment isn’t “How do we automate this journey?” It’s “How do we make sure every part of it is connected, and that the humans at the wheel have what they need when it counts?”
That journey represents something AI, deployed correctly, can support, not replace. The value wasn’t in any single touchpoint. It was in 159 interactions forming a coherent, connected story, with human advisers stepping in when reassurance, context, and judgment mattered most.
The question AI needs to answer in student recruitment isn’t “How do we automate this journey?” It’s “How do we make sure every part of it is connected, and that the humans at the wheel have what they need when it counts?”
Where AI actually delivers
A decade of working in student recruitment across 85 partner universities, managing over 12 million communications in 2025 alone, and supporting 500,000 students to enrolment gives us a clear view of where AI genuinely moves the needle.
The highest-value applications are almost universally back-end: quality assurance at scale, where AI reviews thousands of adviser interactions and surfaces insights no human team could process in-cycle; propensity modelling trained across institutions, which predicts with meaningful accuracy which students will enrol and what interventions will make the difference; and autonomous engagement agents that handle routine enquiries around the clock, within guardrails institutions define, freeing advisers for conversations that require genuine human judgement.
What these applications share is dependence on the depth and breadth of data. A propensity model trained on one institution’s data is almost useless; the student journey is too varied and too long to generate a stable signal in isolation. The same model trained across 85 institutions over a decade of outcomes is a different proposition entirely. This is the data most universities cannot realistically build themselves.
The challenge is not unique to higher education. Airwallex, the global payments platform, has made AI central to its proposition, yet when its CEO was recently interviewed by the BBC, the message he was most emphatic about was: “You can’t vibe-code financial infrastructure.” No amount of AI capability compensates for weak foundations. Data that is fragmented or siloed doesn’t become useful because a more sophisticated tool is placed on top of it.
The right framework for evaluating AI
Before any investment decision, institutions should be asking: what problem are we actually solving? What data does this tool need, and do we have it in a usable state? Is this genuinely AI, or workflow automation with a contemporary label – and if the latter, is that still valuable for what we need?
There’s also an ethical dimension that deserves attention. Any model trained on historical student data will encode historical patterns, including those that could inadvertently deprioritise students with certain characteristics. The organisations getting this right build human oversight in as a structural principle, not a fallback. In higher education, where decisions are not transactional and bias carries real risk, that isn’t optional.
Getting it right
The institutions that will benefit most from AI over the next five years are not those that move fastest. They’re the ones that invest in the right infrastructure first, apply AI where data depth makes it credible, and resist pressure to deploy visible features before the foundations are in place.
The applications that deliver are proven, live, and producing measurable results today. But even institutions that understand this face a structural reality: building infrastructure, training models at scale, and keeping pace with a landscape moving faster than any procurement cycle are not tasks a recruitment team can take on alongside everything else.
Working with partners who have spent a decade building exactly this capacity isn’t about outsourcing ambition. It’s about being realistic about where genuine expertise and data depth sit, and directing institutional energy toward what universities do best.
About the author: Rachel Fletcher is the CEO and co-founder of UniQuest and Keystone Enrolment Services. She has over 20 years of experience partnering with higher education institutions to optimise student engagement and conversion operations in the UK, US, and Australia. Rachel co-founded UniQuest to build a flexible service that could adapt to the rapidly changing needs of educational institutions and students. She has always had a passion for international education and has studied and gained degrees in five different countries.
UniQuest recently launched its next-generation AI-driven student engagement platform designed to transform how universities manage prospective student relationships.
The platform introduces an integrated operating environment for student engagement, combining AI-powered communication tools, predictive analytics and human advisory expertise within a single system.
The company specialises in student engagement, enrolment, and retention services for higher education providers. To date, UniQuest has managed over 70 million student communications, resulting in 550,000 enrolments on behalf of its university partners in the UK, Europe, as well as the US, Australia, and New Zealand through Keystone Enrolment Services.
The post Higher education is investing in AI in student recruitment. The hard part is knowing where it works. appeared first on The PIE News.