Humanising learning analytics in academic advising: how do we balance data and relational advising practice? Free

The use of artificial intelligence (AI) models in the educational domain has seen exponential growth, particularly since 2022 (Baidoo-Anu and Ansah, 2023). AI can support tasks such as materials creation and can tailor learning aids to individual needs, therefore personalising the educational experience of a student (Latrellis et al., 2024). Beyond this personalisation, its responsiveness and speed in addressing student queries have led universities to adopt AI as a supportive tool, including the emerging use of AI within student support (Alwakid et al., 2025) as well as academic advising (Bilquise et al., 2024). Academic Advising often involves the use of learning analytics, which use student data as proxies for engagement and as indicators of learning behaviours (Silvola et al., 2021). Alongside this approach, AI is now being used to analyse patterns in the data, predict student outcomes and make suggestions for targeted learning enhancements (Kumar et al., 2025). In some cases, AI-driven learning analytics is being promoted as a more efficient, consistent and equitable way of personalising learning compared to traditional advising methods (Soomro et al., 2025). Learning analytics can strengthen academic advising by providing data-driven insights that shape meaningful conversations with students, including highlighting learning behaviours students themselves may not recognise (Lowes, 2020). Yet data alone cannot explain a student’s experience, nor can advisors or AI interpret it in isolation. Making sense of data requires collaboration between students and advisors, and institutions need to ensure advisors know how to interpret data, relate to students and develop their competency in humanising data in conversations with students.

The professional body which supports academic advisors in the United Kingdom, UKAT, state that academic advising (or personal tutoring) “personalises learning, promotes student persistence, enhances success, and helps address differential attainment, supporting a student’s academic progress and personal development throughout their time in higher education” (UKAT, 2026). The function of academic support is typically performed by an academic advisor, who is a named member of staff assigned to a student for the duration of their studies (Walsh et al., 2009). Through the academic advisor, a student becomes more connected to an institution and therefore the academic advisor must be able to forge a close connection with a student (Drake, 2011). To achieve this relationship, advisors need to show relatability, empathy and employ an active listening approach to build rapport (Hughey, 2011).

Academic advising has been shown to have a positive impact on student outcomes (Owens, 2015; Erlich and Russ-Eft, 2012) as well as on a student’s experience of university (Drake, 2011). Specifically, the quality of the advisor-student relationship shapes a student’s “psychological contract” with their institution, which directly influences their sense of belonging (Yale, 2019). Because a strong sense of belonging is correlated with increased satisfaction (Palmer et al., 2009) and improved retention (Thomas, 2006), this highlights the importance of the human aspect of academic advising within universities.

Massification of higher education has seen increasingly large student cohorts attend universities in the UK (Boliver, 2010) and, more recently, recruitment cycles have grown ever more competitive as a result of sectoral financial pressures (Rowsell, 2024). Consequently, institutions are facing pressure to monitor student risk relating to both attrition and welfare (PwC, 2023). External financial pressures intersect with internal financial imperatives; in a competitive market, early identification of student disengagement is within an institution’s strategic interest. Alongside this, the Office for Students (OfS) Conditions of Registration require providers to take all reasonable steps to “ensure that all students have access to resources and support to ensure a high quality academic experience, and enable them to succeed in and beyond Higher Education” (OfS, 2022). Increasingly, institutions are turning to metrics to evidence compliance with these expectations.

One emerging approach to predicting student attrition is the use of engagement analytics, where data points act as proxies for student behaviour. Data points might include attendance at scheduled sessions, time spent on a virtual learning platform or online library, or the number of clicks on online resources. Algorithms synthesise these indicators to determine a student’s level of engagement, enabling institutions to identify those deemed at risk of withdrawing. Responsibility for re-engaging these students typically falls either to a central support team or to the academic advisor.

In some more sophisticated Learning Analytics systems, an algorithm is used to make predictions about how successful a student will be in their programme based on their data points (Boroowa and Herodotou, 2022). While predictive approaches have been criticised for the risk of becoming self-fulfilling (King and Mertens, 2023), the absence of such data leaves academic staff reliant solely on what a student chooses to disclose, which depends heavily on the student’s own insight and willingness to share. The availability of student data thus acts as a tangible basis on which to have conversations and offers an enhanced and data-focused approach to advising students (Gutiérrez et al., 2020).

Learning Analytics in academic advising can be understood in terms of a loop (Zanzotto, 2019): student actions generate data, algorithms process that data according to institutional parameters, and advisors then use the outputs to guide conversations with students. In this model, the advisor is positioned as the final point in the loop. Some predictive learning analytics systems include suggestions about how students could maximise their learning based on data alone (Khosravi et al., 2021; Gutiérrez et al., 2020). Yet there remains a gap between the availability of student data and its meaning framed through the lived experience of the student, which can be turned into developmental opportunities that support a student. Without the relational aspects that are critical to effective advising, the advisor risks becoming merely the “human-in-the-loop” of an automated system, interpreting numbers rather than facilitating meaningful learning.

The concept of human-in-the-loop has emerged from conversations about artificial intelligence in education (Hau, 2025). It suggests a one-way system of communication where the Advisor reviews patterns and makes judgements and plans based on data before feeding this back to a student. Khosravi et al. (2021) highlight increasing concern about using predictive models in decision-making tasks without human oversight that affects people. Arguably, an academic advisor oversees the predictive model, but without a deep understanding of data analysis, they can only act as a conveyor of AI predictions (Khosravi et al., 2021). Moreover, academic advising pedagogy that relies heavily on AI predictions undermines the critical value of creating meaning through dynamic conversations with students. Academic advising is typically understood as a form of teaching and learning and learning is “a relational, meaning-making act between people anchored in language, context, curiosity and care” (Hau, 2025). If we consider learning as a one-way loop, we hark back to a Dickensian notion of students as “empty vessels”.

Institutions which invest in learning analytics must go beyond technical training to equip staff not only to interpret data with confidence but to translate those insights into compassionate, human-centred conversations which genuinely support students. The conversations that staff have with students within the academic advising relationship build both meaning and create value. Allowing space for discussion about data enables students to bring meaning to their associated data patterns and reinforces learning analytics as a way to begin a conversation rather than just being a surveillance mechanism. So that deeper, non-superficial, authentic conversations take place, academic advisors need to be supportive, empathetic, encouraging but also purposeful in their approach.

In order to achieve this, universities need to prioritise advisor training, teaching data literacy with ethical framing. Higher education (HE) staff must consistently adopt a holistic view of students, recognising that their academic engagement is shaped by complex, multidimensional lives beyond the classroom. HE institutions have made use of “student personas” which have been created based on the profiles of real students and which incorporate characteristics associated with the 12 specific risks to equality of opportunity identified by the Equality of Opportunity Risk Register (OfS, 2023). Staff are presented with a student persona alongside their fictional engagement and attendance data. Colleagues learn about the student from their profile, consider the data and discuss which aspects need approaching through the advising space. Often, notable parallels are drawn between what we know of a student and what is reflected in the data, reminding staff that data gives a very two-dimensional view of a student rather than the holistic view from having conversations.

As universities adapt to and embrace AI, a central priority must be that academic advising remains a fundamentally human experience. Learning analytics can assist academic advising practice but numbers alone cannot capture the complexities of students’ lives. The value of learning analytics as part of advising, therefore, lies in transforming data into dialogue. Having humanising conversations, where students do not feel judged and where they feel safe to reflect and consider their learning behaviours, will always be of greater value in terms of a student’s experience of university. Institutions need to cultivate a culture of critical data use to encourage advisors to question and contextualise data but they also need training in how to manage conversations with students so that both parties are involved in bringing meaning to the data and that the space created is one of empathy and encouragement. Only then will advisors have the fullest picture of a student and be able to offer the support and developmental guidance that truly meets their needs.

Given all of the above, our following provocation questions are:

1. To what extent should universities use AI tools as a cost-saving measure when it comes to academic advising?

2. How do students perceive AI-assisted academic advising?

3. From a multi-stakeholder perspective: How do we keep the priority focused on humanised academic advising while incorporating data in a rapidly advancing AI world?

4. How can academic advisors develop students beyond just academic support and pastoral care?

5. How can academic advising be consistently delivered across institutions with evergrowing student support needs?