Do Multimodal Learning Analytics Interventions Work? Insights from ISILA Piloting
As part of the ISILA project, one of our key questions has been both simple and ambitious: Do multimodal learning analytics interventions actually work in real educational settings? Following extensive piloting across partner institutions, we can now offer a clear and evidence-based answer—yes, they do, but with important nuances.
What Do We Mean by “Multimodal” Interventions?
In ISILA, multimodal learning analytics interventions combine multiple sources of data and multiple forms of action. These include behavioral data from learning management systems, performance indicators such as assignments and quizzes, and crucially, self-regulated learning (SRL) data capturing motivation, anxiety, and learning strategies.
These insights are brought together through dashboards and translated into both general and personalized interventions, ranging from reminders and additional materials to one-to-one communication and emotional support.
Strong Evidence of Impact
The piloting results show clear positive effects across several dimensions of learning.
First, student engagement increased in almost all courses. Activity levels rose noticeably after interventions, especially when teachers combined analytics insights with direct, personalized communication. In some cases, there were clear spikes in engagement following mid-semester and final-week interventions, showing that timing matters just as much as the intervention itself.
Second, student performance improved. Many students who initially struggled, particularly those who started late or performed poorly early on, were able to recover and improve their outcomes after targeted support. This included higher assignment completion rates, more frequent quiz participation, and measurable grade improvements across courses.
The Power of Personalization
One of the strongest findings from the pilots is that not all interventions are equally effective. General interventions, such as reminders or additional materials, can boost activity, but the most impactful changes were consistently linked to personalized interventions.
Students responded particularly well when teachers reached out directly, acknowledging their individual situation. This sense of being “seen” played a key role in re-engagement, especially for those who were already somewhat active but falling behind.
At the same time, the pilots showed that a subset of students remained disengaged despite repeated efforts, highlighting that even well-designed analytics interventions have limits and may require broader institutional support.
Why SRL Data Makes a Difference
A defining feature of ISILA’s multimodal approach is the integration of self-regulated learning data, and this proved to be a game changer.
Behavioral data alone could show what students were doing, but not why. By incorporating SRL indicators such as anxiety, motivation, and time management, teachers gained a much deeper understanding of student needs.
For example, students with high anxiety responded far better to empathetic, supportive communication than to standard reminders. Similarly, students with low motivation or weak self-management required repeated scaffolding rather than one-off interventions.
This highlights a key takeaway: effective interventions are not just data-driven; they are human-centered.
Timing, Iteration, and Context Matter
Another important insight is that effectiveness depends heavily on when and how often interventions are delivered. The most successful approaches were not isolated actions but part of a structured, iterative process throughout the semester.
Early interventions helped orient students, mid-semester actions supported adjustment, and late-stage interventions proved particularly powerful in re-engaging students and boosting completion rates.
Context also plays a critical role. External factors, such as institutional policies, workload, or even societal disruptions, can significantly influence student engagement and the impact of interventions. In some cases, interventions were only effective once students were able to re-engage due to changing circumstances.
So…Do They Work?
The evidence from ISILA piloting is clear: multimodal learning analytics interventions can significantly enhance engagement, performance, and student support in higher education.
However, their effectiveness depends on several key principles:
- combining multiple data sources, including SRL
- prioritizing personalized communication
- adopting a tiered, iterative approach
- interpreting data within its broader context
Looking Forward
These findings move learning analytics beyond theory into practical, evidence-based teaching innovation. Multimodal interventions are not just effective: they offer a pathway toward more responsive, inclusive, and supportive learning environments.
At ISILA, we will continue refining our approach and sharing insights from practice. The journey does not end with proving that these interventions work; it continues with making them scalable, sustainable, and impactful across diverse educational contexts.