Pacific Union College: Exceeding student expectations with Conversational Search

The challenge

As a small private college competing for students against much larger, higher- resourced universities, Pacific Union College (PUC) faces a pressure familiar across higher education: delivering a digital experience that punches well above its weight.

Prospective students shaped by AI-powered tools in every part of their digital lives arrive at university websites with high expectations – and PUC's existing search experience wasn't meeting them. Analytics showed users frequently searching without finding what they needed, with many bypassing the site altogether and defaulting to Google's AI-enhanced results instead.

For a college whose strength lies in its personal, student-centred approach, that gap mattered. Every unanswered question could lead a prospective student to look elsewhere.

PUC needed a tool that would:

• Answer prospective students' questions instantly, in their own words
• Deliver a more personalised experience without increasing headcount
• Reduce the volume of repetitive enquiries reaching the admissions and financial aid teams
• Reduce failed searches and improve overall search effectiveness across the site

With the Squiz Digital Experience Platform (DXP) already in place, the next step was Conversational Search: a way to give every prospective student an instant, accurate answer in their own words, without adding a single person to the PUC team.

"We need to compete digitally with institutions that have far greater resources. Conversational search gives us a way to do that" – Nic Hubbard, Director of Marketing and Communication, Pacific Union College.

The solution

Conversational Search is an AI-powered search tool that allows users to ask questions in natural language and receive direct, accurate answers drawn from content verified by the institution. Unlike traditional keyword search, it understands user intent, supports multi-step queries, and gives institutions full control over the experience – no external data, no hallucinations.

For PUC, it offered a way to close the gap between what students expected and what the site was delivering – contextual, personalised answers instead of a list of links to navigate alone.

A focused, phased approach

Rather than rolling out across the entire site at once, PUC and Squiz took a targeted approach – starting with the Admissions and Financial Aid sections. These are the highest-stakes areas of the site for student recruitment and conversion, and the most common source of questions from prospective students and their families.

The scoped approach also surfaced something unexpected: outdated content that had slipped through the cracks. Dates referencing 2024 orientation events, scholarship sections missing clear labels, and pages that hadn’t been reviewed in years were all flagged during the content audit and fixed before go-live.

Structuring content for AI required deliberate adjustments too; in the scholarship section, for example, clearly labelling subsections prevented the AI from conflating related topics.

The team worked closely with Squiz throughout – weekly Friday check-ins and a structured rhythm of review and action kept the project moving from kick-off to go-live in just over six weeks. Before launch, Squiz ran multiple rounds of testing and refinement with a 90% answer quality threshold required before the tool could go live.

Critically, this phased approach meant the team finished with something more valuable than a single successful deployment: a repeatable playbook for how to audit, structure, and launch Conversational Search across any section of the site. Every future content slice will be faster and cleaner for it.

The results

The numbers from the first two months tell a consistent story: students are getting answers, and they're not having to ask twice.

Scholarship-related queries dominated activity in the early weeks – prospective students asking questions like "Is there a scholarship available for me?" or "Are there scholarships for visual arts?" Application process questions were also prominent, including how to apply and admissions deadlines.

Conversational Search analytics

Taken together, the two headline metrics point to the same outcome: students are getting complete, useful answers in a single exchange, almost every time.

Follow-up rate

Follow-up question rates stayed low at around 1–2%, suggesting users were largely getting the answer they needed first time.

Success rate

By its second month live, PUC's success rate – the percentage of questions that received a full answer rather than a default response – had reached 97.6%, meaning that almost every student who asked a question left with a direct, useful answer rather than a dead end.

An unexpected benefit: better content

For Nic, the most significant outcome of the project wasn’t the search analytics; it was what the content audit revealed.

“I think the biggest impact was the content audit. It gave us the opportunity to closely look at pages and text that we maybe hadn’t had time to review for a while. I’m proud that we were able to do that, with your help.” Nic Hubbard, Director of Marketing and Communication, Pacific Union College

The process of preparing content for AI surfaced gaps, outdated information, and structural issues that had been invisible to the team for years. The scholarships section was a case in point: content was flagging for vague language, so Squiz worked with the PUC team to optimise it with more specific language, giving both the LLM and prospective students clearer context about the different types of scholarships on offer. That discipline – auditing for clarity, structuring for AI, and keeping content current – is now baked into how PUC manages content across the site.

What's next

Building on the Admissions and Financial Aid rollout, PUC and Squiz are already planning the next content slice. Academics is the frontrunner, with a department-by-department approach under consideration to keep the scope manageable.

The work done so far also puts PUC in a strong position beyond Conversational Search itself. Content that has been audited for clarity, structured for AI comprehension, and kept up to date is better placed for AI discovery more broadly – including the AI-generated summaries and citations that are increasingly shaping how prospective students find information online. For a small institution competing for visibility against much larger universities, that compounds over time.

Longer term, PUC is exploring Squiz Content Intelligence – a capability that analyses how content is performing across the site and surfaces opportunities to improve it, helping institutions make smarter decisions about what to create, update, and retire. Together with Conversational Search, it forms the foundation of a site that doesn't just answer student questions today, but stays optimised for however students search tomorrow.