The University of Greenwich: Transforming student experience with Conversational AI Search

Lorna Hegarty 16 Dec 2025

University of Greenwich

The University of Greenwich is a modern public university in London (UK), known for their diverse, inclusive community, strong research, and unique heritage, with campuses in Greenwich (a World Heritage site), Avery Hill, and Medway, offering a wide range of courses and a vibrant campus life for local and international students.

Industry

Higher Education

Products

The Challenge

The University of Greenwich faced a familiar challenge in higher education: helping prospective students navigate their comprehensive but complex website to find the information they needed.

"Our core audience is Gen Z. They are the early adopters of tech, they know what they like but they’re also really clear about what they don't like as well." – Richard Garratt, Head of Digital Marketing at the University of Greenwich.

With a Gen Z audience that expects modern, intuitive digital experiences, the university recognized several key obstacles:

Information overload: A large website packed with detailed resources, but difficult to navigate for 17-year-olds unfamiliar with university terminology and buzzwords.

Rising expectations: Students accustomed to AI-powered tools on other platforms expected similar functionality from their university.

Competitive pressure: The university was competing for attention not just against other institutions, but against all the brands their audience interacts with daily.

Critical timing: The need to be ready for clearing, a highly competitive and short window of time during which UK universities offer courses to students seeking second-round options.

With clearing fast approaching, Greenwich needed to reduce friction with a solution that could:

Make it easy for students to get the information they need, fast
Allow students to find answers on the website like they would using ChatGPT
Provide search insights so the university could optimize content
Get them ready for clearing within a five-week time frame

The Solution

Working with Squiz, Greenwich implemented Conversational AI Search.

What is Conversational AI Search?

An AI-powered enterprise search tool that allows users to ask questions in natural language and receive direct, accurate answers. It understands user intent and supports multi-step queries to help them get the specific information they need. Critically, it offers the institution full control over the conversational experience. Allowing teams to approve the type of questions answered and content used to develop them. Users ask questions and get instant, accurate AI answers drawn exclusively from content verified by the institution - no external data leakage, no hallucinations.

Beyond a drastically improved search experience for on-site users, Conversational AI Search also supports content optimization for off-site discovery. Auditing tools deliver insights that show how content is interpreted by the LLM driving the experience, so teams can better understand what needs to be done to improve content for external AI search engines.

To implement Conversational AI Search, the university adopted a strategic, phased approach, guided by Squiz Consultants along the way.

Step 1: Setting goals

Squiz and the university’s Marketing and IT teams worked together to define the top three goals:

Enhance the recruitment journey in time for clearing
Improve website content discovery and user experience
Following clearing, free up human support channels

Step 2: Define initial use case

The team followed what Squiz has coined the ‘thin slice' approach. Choosing a smaller ‘slice’ of content to run the tool against, instead of trying to implement it across the whole site at once. This low-risk, high-value approach allowed the team to break the work into manageable chunks – meaning they wouldn’t get overwhelmed and could take learnings from this project and apply them to the next slice.

Greenwich decided to focus initially on university life, support, and wellbeing content.

Step 3: Analyzing content readiness

Squiz supported Greenwich in auditing the key content pages that would support the answers to questions related to the chosen content slice. This involved Greenwich supplying a list of common queries their visitors were likely to ask in relation to their chosen content slice, as well as the content pages that should best support the responses. Squiz then broke the pages down into content fragments to audit the content and determine how useful it would be in answering visitor questions.

Step 4: Getting content AI ready

Based on the results of the content audit, Squiz provided Greenwich with a series of recommendations that included simple tweaks to key pages alongside some brand-new content pages to support great answers.

The added benefit of having these content insights is that Greenwich has also been able to optimize it for LLMs off their site. Greenwich leveraged the Conversational AI Search LLM on their site in combination with expert insights to understand how to get their content ready for the implementation. By optimizing content for their on-site LLM, they also enhanced content discoverability for the university via external AI search engines, where many are now starting their research.

Step 5: Conversational AI Search implementation

Once Greenwich was ready to start the implementation, the team followed two phases:

Backend setup: Squiz wrote the initial use case within the platform – this meant ensuring that when users asked questions, the correct content piece was selected for the answer.

Alongside creating the initial use case, Squiz and Greenwich provided guidelines to the Conversational AI Search chatbot. These guidelines were designed to ensure the chatbot would sound authentically Greenwich – reflecting the university's distinctive tone of voice while staying within defined content boundaries.

This authenticity extended beyond just how answers sounded. At the heart of Greenwich's approach was their commitment to "education without boundaries" – an ethos that drives everything they do. Implementing Conversational AI Search was a natural extension of this philosophy: removing barriers in the discovery and application journey, making information accessible exactly when and how prospective students and applicants need it, without the friction of complex navigation or university jargon.

Frontend implementation: To make the Conversational AI Search function easy to use, Squiz and Greenwich created a new search page that combined the two types of search – the traditional keyword search bar supplemented with new Conversational AI Search technology.

By combining the two types of search, whatever students search for, they will get a result. It’s a topic the Conversational Search LLM can support, the response appears at the top of the page. Below that, students can still see the regular keyword search results they're used to.

This new search page was built and tested in parallel alongside Greenwich’s existing search page, so when the time came, it would be easier to switch the pages over seamlessly.

Step 6: User testing

Squiz consultants developed a framework for assessing the ‘acceptability’ of answers. In this testing process Squiz generated questions to run through the Conversational AI Search chatbot, before sending validation the answers with Greenwich.

The framework ranked answers across four key pillars:

Accuracy
Completeness
Scope
Tone

For any answers that didn't meet the acceptability threshold, Greenwich provided feedback on why, allowing Squiz to investigate and provide recommendations – whether that meant fixing content, adjusting use case parameters, or refining answer guidelines. This process involved multiple rounds to refine the content and technology.

Once a threshold of acceptability of answers was hit, a final phase of testing was carried out by a panel of Greenwich testers with the brief: ‘try and break it!’

Step 7: Launch

On launch day, Conversational Search was made live by simply switching their old search page URL with the new one that included Conversational AI Search.

Students could then interact with the conversational search experience from anywhere on the site via the main site search icon. Greenwich also introduced the new search feature via a content card component on their Clearing landing page.

Greenwich made sure to communicate to users that they were trialing AI search, maintaining transparency about the technology.

In the first month, Greenwich saw 4,821 questions asked using the Conversational Search feature, giving them valuable insights into the information visitors were most commonly looking for on the site.

Step 8: Continuous monitoring and optimization

By monitoring internal dashboards, Squiz can provide ongoing support by tracking usage analytics and conversation performance.

Alongside Squiz support, Greenwich has full access to their chatbot console, where they can review all conversation history, monitor generated answers, and identify any responses that need adjustment or content that requires updating.

This dual approach enables both strategic optimization support from Squiz and day-to-day monitoring capabilities for Greenwich's team.

The Outcomes

The implementation project went from kick off to launch in just five weeks and, within the first month, operating on just their first slice of content, the tool was already delivering great results:

Improved user experience creating more engaged students

4,821 questions were asked
75% of questions resulted in a conversational response
Users were asking, on average, 2.45 questions per session
Users spent an average of almost four minutes engaging with the tool

Greater internal efficiency and content discoverability:

Beyond improving the experience for users and the speed with which they can now find the information they need, the implementation of Conversational AI Search has also given Greenwich much greater, richer insights into what people want to know and, critically, whether their content is delivering the right answers.

They have gained:

Insights into student interests – with popular topics including accommodation, facilities, student services, mental health and fitness
Identification of content gaps – information on what students were looking for that wasn't easily discoverable on the site

“Conversational Search gave us a new way to connect with future students. It delivered fast, accurate answers to over 5,000 users during Clearing — and the insights we’re now gaining are reshaping how we create and refine our content.” Richard Garratt

SEO and AI search optimization – in preparing their content for Conversational AI Search - which is powered by an LLM on their site - Greenwich has also optimized it for LLMs off their site.
Insights to support student retention - given the types and conversational formats of the questions students are asking on the site, Greenwich has a much better idea of the information current students need to improve their university experience.

The future

Greenwich's vision for Conversational AI Search extends well beyond the initial implementation:

Content expansion: Plans to gradually include course information, working through the complexity of 400+ frequently changing courses while using the process to improve underlying content quality.

SEO integration: Leveraging Conversational AI Search learning to enhance their presence in AI-powered search tools, recognizing that AI tools are increasingly driving search behavior.

Continuous improvement: Using real user data to identify and fill content gaps, creating a better overall experience for prospective students.

AI search is now a competitive differentiator in clearing, student recruitment and beyond. Greenwich is proving how fast an institution can move and how quickly students will adopt a new experience when it is truly intuitive and aligned to their expectations and behaviours.