Hi, this is Ray.

I should probably disclose something upfront before writing about this topic, in the spirit of intellectual honesty: I've spent a substantial chunk of my professional life building online language courses. I have a vested interest in the question of whether online learning works. You should probably read this newsletter knowing that.

That said, I'm going to tell you something that might be surprising given the disclosure. Most online courses, including some I've built and many I've bought, don't work as well as they could. The research backs this up. The completion rates are humbling. The actual learning outcomes are wildly variable. The marketing for online courses tends to oversell. And the honest picture of what online courses can and can't do is more nuanced than either the enthusiasts or the critics typically present.

I'm writing this newsletter as someone who genuinely believes online learning is one of the most important educational developments of our era AND someone who has watched too many people buy too many courses they never finished. Both things are true. The technology has democratized access to instruction in ways that genuinely change lives. Most people who buy the courses, including me at various points, fail to extract most of the value from them.

Today's newsletter is about that gap. What the research actually shows about online courses, why so many learners abandon them, and the specific things you can do differently to actually finish (and learn from) the next course you take. Let's get into it.

From our partners at Gauntlet AI:

AI is already handling large parts of execution. That shift is not coming later. It is happening now.

What is left, and becoming more valuable, is the ability to design systems, apply AI thoughtfully, and own outcomes in production. That is the work strong teams expect from senior engineers in 2026.

Gauntlet is built for engineers who want to operate at that level. In a single week, 15 hiring partners conducted 246 interviews with challengers onsite in Austin. Apply now.

Apply to Cohort 6

_{Must be a US citizen to qualify.}

Let me start with what the research actually shows, because the picture is more interesting than the headlines suggest.

The good news first. Multiple studies have found that well-designed online courses can produce learning outcomes comparable to traditional classroom instruction. According to one analysis, Jordan, Lacerda, and Wells conducted a meta-analysis that found MOOCs to be generally as effective as traditional courses in terms of knowledge acquisition, with availability of high-quality resources including videos, quizzes, and discussion forums contributing to successful learning outcomes. When researchers compare what people actually learn from well-designed online courses to what people learn in equivalent classroom settings, the outcomes are often roughly equivalent. The medium itself isn't the problem.

This is a genuinely important finding. For a long time, the assumption was that "real" learning required in-person instruction, and that online learning was a watered-down substitute. The research doesn't support that. The same knowledge can be effectively transferred through well-designed online instruction. The format isn't what's limiting. Something else is.

What is the something else? Here's where the research gets less flattering. The same body of literature consistently identifies a brutal truth about online learning: completion rates are catastrophically low. Most MOOCs see completion rates in the single digits… often under 10% of enrolled learners actually finish the course they signed up for. As one analysis pointed out, key concerns include low completion rates observed in many MOOCs, with various factors contributing to this issue, such as lack of personalization, absence of accountability, and limited interaction with instructors. Most people who sign up for online courses don't finish them. That's the central problem of online learning, and it's worth taking seriously.

So the honest synthesis is this: online courses can be as effective as traditional courses for the people who actually complete and engage with them. The challenge is that most people don't complete or engage with them. The medium works. The problem is the learner-medium relationship.

Let me get specific about why the abandonment rates are so high, because understanding the failure modes is the first step toward avoiding them.

According to research on MOOC effectiveness, the satisfaction level of the learner is affected positively by variables like online self-regulated learning which includes goal setting, behavioural variables and perceived course usability. Translation: the learners who succeed in online courses are the ones who can self-regulate… set their own goals, manage their own time, monitor their own progress, push through difficult parts without external pressure. The learners who don't have these skills tend to abandon online courses, regardless of how good the courses themselves are.

This matters because traditional in-person education has been doing a lot of the self-regulation work FOR students for their entire lives. The teacher set the schedule. The class set the pace. The grade requirement created the accountability. The physical presence of other students created social pressure to keep up. When you transition to online learning, all of that external scaffolding disappears. Suddenly you're responsible for everything that the institutional structure used to handle. Most people don't have the self-regulation capacity to immediately step into that role, because they've never had to develop it.

A 2025 study analyzing 8,602 MOOCs found another important pattern. According to the researchers, course quality, content richness and interactivity significantly influence course dropout rates and pass rates, with more interaction not always being better… excessive content and online interaction can actually hinder learning effectiveness. The intuition that more content equals more value turns out to be wrong. Many courses pack so much in that learners drown. The optimal courses have focused content, clear structure, and appropriate interactivity… not maximum content and maximum interactivity. If you've ever bought a $497 course with 47 hours of video and quietly abandoned it after the first week, this finding may help explain why.

The other common failure mode is the goal mismatch. People sign up for courses with vague aspirations… "I want to learn data science," "I should improve my Spanish," "everyone says I need to know AI." Vague goals produce vague engagement, which produces abandonment when the work gets hard. The learners who succeed in online courses almost always have specific, concrete reasons for taking them… a project they want to build, a job application they're preparing for, a problem they actually need to solve.

Despite the completion rate problem, online courses have genuine strengths that other learning modalities don't quite match. Let me name them, because they help clarify when online courses are the right tool.

Access to instruction you couldn't otherwise get. This is the foundational benefit. Online courses give you access to teachers, content, and curricula you literally couldn't access any other way. Top universities make courses available globally. Specialists in narrow fields can teach learners they'd never have met. People in remote areas, with disabilities, or with constrained schedules can learn things they couldn't learn through in-person instruction. This benefit alone justifies the existence of online learning as a category, even with the completion rate problem.

Self-paced progression. Online courses let you go at your own speed. Slow down for things you find hard. Skip ahead through things you already know. Re-watch the explanation that didn't quite click the first time. The match between learner pace and material can be much tighter than in a classroom, where everyone moves at the average pace whether or not it fits any specific student.

Cost efficiency. For most people, online courses cost less than equivalent in-person instruction. Often dramatically less. Free MOOCs from major universities, $20 Udemy courses, $50/month subscriptions to comprehensive platforms… the cost has come down enormously. The economic accessibility of learning has expanded in genuinely important ways.

Just-in-time learning. When you need to learn something specific for a project, a job, a problem, online courses can deliver targeted learning faster than enrolling in a traditional class. You can find a course on the specific topic, work through it in days rather than semesters, and apply what you learned immediately. This kind of flexible, targeted learning was much harder before online courses existed.

The ability to revisit. Unlike a lecture you attend once, a recorded course can be returned to. Forgot how something worked six months later? Re-watch the relevant lesson. The persistent availability of the material means the course is a resource you keep, not just an experience you have.

Honest section. Online courses aren't equally good for everything, and matching the medium to the goal matters.

Skills that require real-time feedback. Some learning genuinely benefits from immediate human feedback that responds to what you specifically do. Public speaking. Certain physical skills. Complex negotiation. Anything where the value is in someone watching you do the thing and responding in real time. Pre-recorded online courses can teach you about these skills, but they can't replace the back-and-forth with a real teacher. For these domains, online courses are usually a starting point, not a complete solution.

Highly social learning. Some learning is fundamentally social… the questions that come up in discussion, the perspectives other students bring, the spontaneous conversations that follow class. Online courses with discussion forums attempt to provide some of this, but they don't usually match what happens in a vibrant classroom or study group. If the social dimension is core to what you're trying to learn, online courses alone will fall short.

Domains with hidden curricula. Some fields have important informal knowledge that doesn't show up in any formal curriculum… the implicit standards, the cultural norms, the unstated expectations. This stuff usually transmits through apprenticeship-style contact with practitioners. Online courses, which by their nature systematize the formal curriculum, often miss the informal layer entirely.

Learners with low self-regulation. If you genuinely struggle to manage your own time, motivation, and follow-through, online courses will mostly amplify that struggle. The lack of external scaffolding makes the self-regulation problem worse, not better. For these learners, a structured in-person environment may be a better starting point until self-regulation skills develop.

Okay, the practical part. If you've abandoned online courses in the past and want to do better with the next one, here's what the research and experience suggest.

Pick courses with specific personal motivation. Don't enroll in things because you "should." Enroll because you have a specific reason that connects to something you actually care about doing. The Spanish course because you have a trip planned in eight months. The Python course because you have a specific project you want to build. The marketing course because you have a real product you need to launch. The specific motivation provides the durability that vague aspirations don't.

Start with shorter, more focused courses. A 4-hour course you finish is more valuable than a 40-hour course you abandon after week two. Especially when you're rebuilding your online learning habits, picking shorter courses lets you build the muscle of actually completing things. Once you've completed a few short ones, you can take on longer commitments with realistic expectations.

Schedule it like a class. "I'll work on the course when I have time" produces abandonment. "Tuesdays and Thursdays at 7pm, I do my course work for 45 minutes" produces completion. The schedule is doing the work that institutional structure used to do. Without the schedule, you're trying to push against your own inertia every single day. With it, you've automated the decision.

Create artificial accountability. Since the course doesn't provide accountability, you have to manufacture some. Tell people you're taking the course and ask them to check in. Sign up with a friend and hold each other accountable. Join the course's Discord or forum if there is one and actually participate. Find external pressure to substitute for the institutional pressure you used to have.

Engage actively, don't just consume. Watching videos isn't learning. Doing the exercises, taking the quizzes, building the projects… that's where the learning happens. As one analysis noted, learner motivation and self-regulation skills play a crucial role in MOOCs, as learners must actively engage with the course material without direct supervision. The active engagement isn't optional. It's the difference between a course that produced learning and a course that just produced screen time.

Apply what you learn immediately. The fastest way to forget course material is to learn it and never use it. The fastest way to retain course material is to start applying it the moment you've learned the foundations. If you're learning marketing, run a small campaign while you're in the course. If you're learning Spanish, start having conversations from week one even if they're terrible. Application multiplies retention.

Don't enroll in too many at once. The pattern of buying five courses and starting all of them is one of the most reliable ways to finish none of them. Pick one. Finish it. Then pick the next one. The single-tasking approach produces dramatically more completions than the multi-tasking approach.

Quit gracefully when needed. Sometimes a course genuinely isn't a good fit. The instruction is unclear, the pacing is wrong, the material isn't what you expected. Quitting an unfit course is not the same as the abandonment pattern. It's a deliberate choice to redirect to something more useful. The skill is in distinguishing between "this isn't working because something is wrong with the course" and "this isn't working because I haven't shown up consistently." Different problems. Different responses.

Here's what I want you to take from all this. Online courses are genuinely powerful learning tools. They've made instruction accessible at a scale that would have been unimaginable two decades ago. People who use them well learn things they couldn't otherwise learn. The medium itself works.

The challenge is that most people don't use them well. They buy them with vague motivation, schedule them loosely, engage passively, and abandon them when life gets in the way. The result is an enormous gap between the potential of online learning and the actual learning that most people extract from it. Closing that gap is mostly about learner behavior, not about course quality.

If you've been buying courses you don't finish, the problem isn't you and it isn't the courses. It's the relationship between you and the medium. The relationship can be improved with the specific practices I've described. The next course you take can be the one you finish, if you set it up differently from the courses you've abandoned. The pattern can change.

The democratization of learning is real. The cost of access has fallen dramatically. The quality of available instruction has risen substantially. But none of that produces learning unless you actually do the work, on a sustained schedule, with active engagement, until you finish. The medium is necessary but not sufficient. The other half is you.

Pick the next course carefully. Schedule it like it matters. Engage actively. Apply what you learn. Finish it. The version of you that does these things, consistently, has access to almost any field of knowledge at almost any time, for very little money. That's a remarkable thing. Use it.

Even Bilbo had to read maps. The Shire was small, but the world was large, and knowing the world meant studying it. Your maps are online now. They cost very little. They're waiting. The question isn't whether the maps are good. The question is whether you'll actually study them.

Keep learning (and keep finishing courses),

Ray