In this blog, I discuss the learning analytics project I ran across three undergraduate economics courses, over two semesters. In the fall semester, I taught Microeconomics 1 to first-year Economics and Data Science students. In the following semester, I taught Introduction to Economics to first-year and Introduction to Macroeconomics to second-year students, both groups were from the BSc Industrial Management programme. In each module, students were split into a treatment group and a control group. In the following, I will outline what worked well, the challenges, and what I would change going forward.

What Worked: Conditional Access Quizzes

The best part of the design was the conditional access quiz. Each week, students had to either pass a quiz or try it three times before they could move on to the next set of materials. This gave the module a clear rhythm. It did more than just track who was engaged; it helped guide students toward better study habits.

In a large class, it is simply not possible to track every student closely on your own. This is exactly where learning analytics helps. It gives us the chance to observe each student's progress in a way that would not be possible otherwise, and to step in early when someone starts to fall behind.

I also added weekly lecture notes, slides, and a "question of the week" experiential polling activity. Together, these were meant to do more than measure engagement. They were meant to show students how to study, and to keep them motivated all semester, not just before exams. I think, this combination worked well. Students responded to having a clear path to follow.

The Challenges

The intervention paid off, but it also came with some challenges.

• Building the platform. HERO is the learning management platform used at our university, where course materials, quizzes, and student activity are managed online. The platform did not havea quiz feature. I had to build a custom setup on the HERO platform to support conditional access, different from the standard setup used elsewhere. This took almost three months to design and prepare, before the quizzes were even ready to use.

• Building the quizzes. Each quiz had to be built one by one. First, I created a bank of questions, then picked from that bank to build each week's online quiz. I did the same for the "question of the week" activity. Since I had a treatment and a control group, I had to repeat every step twice. This doubled the workload.

• Tracking performance on time. Keeping up with the "question of the week" and student performance was hard given the pace of a live module. HERO also did not show which questions students got wrong. To give students feedback, I had to take a screenshot of each wrong answer and send it to them. This was slow and time-consuming in a large class.

• Time for closer tracking. Because building the platform, writing the quizzes, and giving manual feedback took so much time, I could not follow each student as closely as I wanted. I sent email reminders and warnings, but with more time, I believe this could have been more consistent and firmer.

• Technical issues. I ran into recurring technical problems, including questions that sometimes failed to upload, which added more friction to an already slow process.

Where I See Room for Improvement

Looking back, three things stand out for next time:

• A better platform. A system that supports conditional access by default, and shows which answers were wrong, would save a lot of manual work. That time could go toward what matters most: talking to students.

• Faster quiz building. A workflow that cuts down repeated work between the treatment and control groups, for example through templates or bulk creation tools, would save a lot of preparation time.

• More consistent follow-up. With less administrative work, I would like to build a steadier routine for reaching out to students who fall behind, rather than fitting it in around everything else.

What Students Said

Student feedback, gathered across all three courses, offers a valuable check on whether the design was actually achieving its intended purpose.

Several students spoke to the sense of responsibility and self-tracking the structure encouraged. One Economics and Data Science freshman wrote:

"I just want to thank our lecturer for this class. After the online quizzes, my sense of responsibility increased, and I started to understand the meaning of this course and its concepts much better. I hope that after the midterm, I did improve my responsibility."

Another described how it helped them stay on top of their own learning:

"Learning analytics helped me understand my progress throughout the course. It showed me which topics I understood well and which ones I needed to review. It made my learning more organized."

Students in my Introduction to Macroeconomics course, taught to Industrial Management sophomores, echoed the same theme, particularly around staying on track:

"Weekly quizzes really helped me to prevent falling behind. Sometimes, students need someone to remind them that they should stay on track."

A freshman in Introduction to Economics, also from Industrial Management, spoke to how the combination of materials worked together:

"All the data provided was useful. After going through lecture notes, doing quizzes was a very effective way to learn the concept. Thank you!"

Others reflected on how the monitoring itself felt from their side, which matters just as much as whether it worked:

"Overall, I felt comfortable with the use of engagement monitoring in this course. The process seemed professional and respectful of my privacy."

Across three different courses and three different student cohorts (first-year Economics and Data Science and Industrial Management, and second-year Industrial Management), the same pattern kept surfacing: structure, paired with a well-timed nudge, makes a real difference to how students engage, and students themselves recognize and value it.

Closing Thoughts

Despite the workload and the platform issues, I still believe, structured required engagement activities such as conditional access quizzes, paired with clear weekly guidance, make a real difference to how students engage with learning materials. The lesson here is not that the approach was wrong, it is that the tools behind it need to catch up with the teaching idea itself.

This is where I see the real value of a learning analytics framework. In a large class, it is simply not possible to track every student closely on your own, but learning analytics gives us the chance to observe each student's progress and step in early when someone starts to fall behind. This matters most for first-year students, who often do not yet know how to study or what to prioritise each week. They need a structure that shows them the way, not just data that tracks them from a distance. I believe this is exactly where the responsibility falls on us as academics: to use the tools we have to guide students, not only to observe them.