You have a Udemy account with eleven courses in it. Three of them you bought on sale for $12.99 and never opened. Four of them you made it through the first two modules and then something at work happened and you lost the thread. Two of them you finished, technically, but if someone asked you to explain what you learned, you’d stall. The remaining two you’re pretty sure you bought twice because you forgot you already owned them.
Somewhere on your bookmarks bar is a Coursera specialization you enrolled in when you decided you were going to learn data science. It’s been there for fourteen months. The progress bar reads 3%.
You feel slightly bad about this. You shouldn’t. Not because you don’t care about learning — you clearly do, you keep buying the courses — but because the format you’ve been trusting to teach you things has a structural problem that has nothing to do with your motivation.
The completion rate tells the story
Across Coursera, edX, Udemy, and similar platforms, average MOOC (Massive Open Online Course) completion rates sit at around 5% — and that’s among students who signed up intending to complete the course. Researchers at MIT and Harvard studied millions of edX enrollments and found that only about 5% of enrolled learners ever earned a certificate. Reich and Ruipérez-Valiente (2019), examining over a decade of MOOC data, concluded that structural problems with the format — not learner motivation — explained most of the dropout.
Sit with that for a second. If you enrolled in an in-person class and 95 out of 100 students dropped before finishing, you wouldn’t conclude that all 95 people lacked discipline. You’d ask what was wrong with the class. Somehow, when it’s online courses, the narrative flips: the platform is fine, the students are the problem.
They’re not the problem. The format has three design failures that compound each other.
The first is passive consumption at the wrong time of day. Most people open course videos in the evening, after work, in the same posture they use to watch TV. Passive video at low-attention hours produces almost no encoding. You can watch a 45-minute lecture on SQL joins and retain essentially nothing, not because you weren’t trying, but because your brain is not in encoding mode at 9pm on a couch.
The second is the mismatch between what feels like learning and what actually is. Watching an expert explain something feels productive. It’s comfortable, it’s frictionless, it generates a vague sense of progress. But passive reception is not the mechanism by which skills transfer. Practice, application, and retrieval are. A course that’s 80% lecture and 20% exercise has those proportions backwards.
The third is the lack of a forcing function. An in-person class has a scheduled time, a room you have to walk to, an instructor who notices if you’re not there, and classmates whose presence creates mild social pressure. An online course has a bookmark. The course is not competing on equal terms with every other claim on your attention at 9pm. It’s going to lose.
What active learning research actually says
The distinction between passive and active learning has been studied extensively since at least the 1960s, but the modern framing comes from cognitive psychology. Active learning — anything that requires you to retrieve, apply, or produce, rather than just receive — consistently produces better retention than equivalent time spent in passive modes.
This isn’t subtle. Research on the testing effect (covered more directly in the remember what you read article) shows that retrieval practice dramatically outperforms re-reading and re-watching. When you’re a student, the instinct is to review your notes again — it feels rigorous. What actually consolidates learning is closing your notes and trying to recall what they said.
For skill-based online courses, the application is direct: the exercise at the end of a module matters more than the lecture. The project is more important than the walkthrough. The 20 minutes of doing produces more than 60 minutes of watching — especially when the watching happens at low attention.
This means the structure of most online courses is approximately backward for your schedule. The passive content (which you’d absorb best when alert and undistracted) is presented first, in the longest chunks, at the time of day you’re least likely to schedule focused attention. The active content (which you could actually do at lower attention if you’d already understood the concept) comes at the end, as an afterthought, after you’ve already run out of energy.
Tonight: pick the one course and schedule the exercise
There’s no value in optimizing across your whole backlog. The question tonight is simpler: which one course has the highest pull for you right now? Not “which one would be most useful for my career” or “which one cost me the most money.” Which one do you actually want to learn?
Open that course. Scroll past the introductory lecture — you’ve probably seen it. Find the first hands-on exercise, lab, or project. That’s what you’re doing tomorrow.
Now: when tomorrow is when you’d normally open TikTok or your news app on your phone? That slot — even if it’s only 15 minutes — is more valuable for learning than an hour of evening video-watching. The course doesn’t require a dedicated learning environment. It requires a specific, recurring time that you actually hold.
Schedule 15 minutes tomorrow, same time each day, with a specific label: “[Course name] — exercise only.” If the course has video instructions for the exercise, fine. But the goal of the time block is to do something, not to watch something.
One specific thing that helps: download whatever you need to do the exercise before the time block, not during it. If you need to install software, set up an account, or download a dataset, do that tonight. Decision fatigue at the start of your practice block is the most reliable way to spend the 15 minutes on setup and nothing else.
A personal observation: the most useful course I’ve ever taken had almost no video. It was a series of exercises with written explanations, and it took about 20 minutes a day. I finished it in three weeks. The most useless course I’ve taken was 40 hours of beautifully produced video that I watched in fragments over six months and cannot describe to you today. The production quality of the lecture has zero correlation with how much you learn from it.
Your 14 unfinished courses aren’t evidence that you’re bad at learning. They’re evidence that you’ve been handed a format that’s designed for completion optics, not actual skill transfer. Pick one, skip the video, do one exercise tomorrow at the same time you’d otherwise scroll. That’s the rep that matters.
For getting more out of what you already consume — including books, not just courses — what to do after you read a chapter uses the same retrieval-practice principle in a format you can run in three minutes.