AI is flowing into our universities, whether we like it or not.
Yet there is an immediate and actionable opportunity AI provides that most institutions are missing – using AI to actually hear students better and act on their feedback faster than the traditional annual survey cycle permits.
Most institutions are sitting on thousands of open-text comments that remain unread or under-utilised. While quantitative scores are scrutinised, the qualitative “why” behind those numbers is often lost to the limits of manual analysis.
All the things you said
When a survey closes, quantitative results move to dashboards, but thousands of student comments often sit in the system, partially read or set aside until the next cycle begins. This is rarely a failure of institutional culture; it is a problem of capacity and prioritisation.
Most institutions still handle open comments manually, often via inconsistent Excel exports that vary by faculty. This results in no longitudinal view, no faculty comparison, and no ability to cut data by student demographics or compare to others in the sector. Feedback becomes a lag indicator, informing retrospective reports rather than driving real-time, in-year improvements.
AI does not fix the culture around student voice, but it helps address resource-based challenges which limit its full potential.
An alternative solution
At King’s, we have sought to move beyond surface-level readings of the National Student Survey and to engage more deeply with what our students are telling us about their educational experience, using deeper insight to inform decision-making rather than treating NSS outcomes as an end in themselves.
To that end, we have applied AI-powered analytical approaches to our NSS free text data to test whether we are hearing students’ voices consistently across faculties and disciplines and – crucially, for the first time – across demographic groups, to examine whether institutional narratives about the student experience are genuinely borne out by lived experience. We are also able to see how the sentiment of our students compares to others in the sector using the technology. The margins are fine.
The scale and pace of this work have been genuinely transformative. In 2025, more than 1,700 individual student comments were processed within two weeks of survey publication, generating over 5,000 category combinations and enabling us to move rapidly from headline metrics to evidence-informed dialogue within an increasingly compressed summer decision-making window.
This analysis is making a tangible contribution to the evaluation of strategic and transformational activity, allowing us to track in near real time how changes which represent a significant and sustained investment of effort by colleagues across the institution are being experienced by students and having an effect. A particular area of focus has been assessment and feedback, where considerable work has been undertaken to reform assessment systems and processes across the university. This has included streamlining complex workflows, reducing duplication, and removing multiple, often manual, steps within the assessment lifecycle to ensure greater consistency and reliability and, critically, the timely return of marks to students.
Since the work began, we have seen a 5.6 per cent improvement in the assessment and feedback metric. Through the addition of sentiment and thematic analysis, we are now able to contextualise this improvement and further evidence impact through a 7.1 per cent reduction in negative sentiment relating to assessment and a 10.3 per cent narrowing of the assessment sentiment gap. Together, these outcomes reinforce confidence that sustained, institution-wide reform is having its intended effect and, critically, that the scale of change undertaken is both justified and worthwhile.
The analysis is also supporting and strengthening the evidence base for developments in campus space, digital education and extra-curricular provision, ensuring that investment decisions are grounded in a robust understanding of student priorities and experience.
Beyond the numbers, this work is providing senior leaders with reassurance that efforts to close longstanding satisfaction gaps are not only well intentioned but have demonstrable impact, helping to shift the culture around NSS performance away from compliance and towards collective learning and enhancement.
Equally significant is the way this approach has enabled us to reach a wider internal audience, including professional service and cross-cutting teams who may not be at the coalface of educational delivery, but whose work nonetheless shapes the conditions in which education is experienced.
By making insight more accessible, interpretable and relevant, we have fostered a stronger sense of shared ownership of student feedback, building a broader institutional community around NSS intelligence and reinforcing the principle that high-quality education is a collective endeavour.
A different view
This is not a single-institution story. At Edinburgh Napier University, NSS verbatim comment analysis previously involved individual coding in Excel, using a coding frame reflecting areas of strategic priority within school action plans. The process was time-consuming and resource intensive.
In 2025, Edinburgh Napier analysed 1,701 NSS comments using Student Voice AI via evasys. Ninety-eight per cent of substantive comments were allocated at least one category, producing 5,340 comment-category combinations. Results were shared with schools within one to two weeks of NSS results being published – a process that previously took five to six working days of manual work.
The comparison between key themes from manual coding and Student Voice AI produced a very high degree of similarity, identifying the same themes and areas of focus at school level. This analysis also offered the ability to analyse multiple different groupings with minimal timing impact, and a coding frame split by a larger number of categories than was possible with manual coding, allowing themes to be identified at a more granular level.
Crucially, it identified specific areas of action which helped schools understand what was behind the quantitative ratings they had received. These insights fed directly into student success action plans produced by each school, supporting the enhancement of the student experience.
As Nicola Kivlichan, head of market and student intelligence at Edinburgh Napier, noted, examples of actions for 2025/26 student success action plans were identified from comment analysis. These included addressing late changes to timetabled classes, perceptions of changes to programme structures as a result of a school merger, concerns over the use of groupwork in the final year of study (leading to research to support enhancement of groupwork), ensuring out-of-office messages identify other members of staff for students to contact, and identifying where students want more practical experience on their courses.
Start with the problem, not the tool
At King’s, the work began with a desire to understand our quantitative data better through text comments, to harmonise and reduce time spent on analytical practice, affording more time for understanding and action, and to use robust management tools for effective executive reporting. It did not start with a procurement decision. The technology must serve the inquiry.
Keeping humans in the loop remains vital. While AI categorises and identifies sentiment, designated staff must read and validate summaries before they reach senior leadership. Student Voice AI operates supervised machine learning models, trained using human-labelled comments. The model is supplied with hundreds of thousands of manually labelled sentence-category pairs. At evasys we recommend stakeholders read and validate AI summaries before action. That human-in-the-loop step balances speed and safety.
Augment, don’t replace. The goal is to give professionals specific, timely data to act upon – turbo-charging their capability rather than de-skilling teams.
Direction of travel
The ambition shold be to connect the survey ecosystem, module evaluations, in-year surveys, and longitudinal trends, into a joined-up feedback system. The sector has spent years asking students what they think. AI makes it possible to hear the answer and act before the next cycle begins.
In a volatile higher education environment, institutions must steer their AI adoption with clear intent. The iterative, evidence-driven cycle is how you get rapid, actionable insight without sacrificing the human relationships students value or risking harm to vulnerable groups. Go forth and steer the flow with intent – don’t just drift with it.
This year, Wonkhe and evasys have convened a community of institutional leaders working on student feedback to build a new way of thinking about student feedback systems. If you’re interested in hearing what we’ve come up with, join us for a free online webinar on Tuesday 16 June 12-1.30pm. Sign up here.
Should this be labelled as sponsored content? It is essentially an advert for evasys, no?