The share of courses students complete rose from 41% to 58%

Context and challenge

The online learning platform was growing in sign-ups, but students were dropping courses halfway through: fewer than half made it to the end. Course designers gathered progress data by hand from several systems — preparing a report on a single cohort took up to two days. Product decisions were made on gut feeling, without metrics.

Solution and the team’s role

Analysis. We walked the student journey from sign-up to certificate and marked the points where drop-off happens most often. It turned out that most of the losses were not down to “difficult material” but to organisational gaps: students did not get feedback on time and lost their rhythm.

Design. We designed an engagement metrics system — completion, regularity of study, time to first feedback — and a funnel that made these metrics visible in real time.

Development. We automated the routine of the learning process: reminders based on each student’s schedule, automatic allocation of assignments for marking, and progress dashboards for course designers instead of manual data exports.

Launch and training. Changes were rolled out in iterations with A/B testing. The course design team learned to read the metrics and run experiments without developers — and that became the main long-term effect.

What we changed after launch

The first version of the reminders proved too intrusive: some students switched notifications off entirely. We rebuilt the logic around “rarely, but to the point” — reminders were tied to each student’s personal study rhythm rather than the course calendar. Students stopped opting out of notifications, with no loss of effect.