State of Online Course Creation 2026: Data from 32,000+ Courses

In 14 years of running Ruzuku, I've watched more than 32,000 courses launch. Some became six-figure businesses. Most didn't. The online course industry crossed $325 billion in 2025, and yet the dominant narrative is crisis: completion rates are catastrophic, AI is eating the middle, the gold rush is over. Both stories are true. This report is about what separates the creators who thrive from the ones who stall.

My PhD research at UNC-Chapel Hill focused on human-computer interaction: how people learn and change behavior through technology. I co-founded Ruzuku in 2011, and in the 14 years since, I've watched more than 32,000 courses launch on the platform. I wrote The Business of Courses (Mirasee Press, 2021) to capture the patterns I kept seeing, patterns that most industry commentary either missed entirely or reduced to marketing platitudes. The patterns in this data are not random. They reflect predictable forces: the psychology of learning, the economics of attention, and the structural choices creators make when they design their programs.

Here are six findings from that analysis. Some will confirm what experienced creators already suspect. Others will challenge assumptions the industry has repeated so often they've hardened into received wisdom.

Chapter 1: How Big Is Online Course Creation in 2026?

The global eLearning market reached approximately $325 billion in 2025, with projections from Statista¹ and IMARC Group pointing to $365–400 billion by 2026. The MOOC segment alone — platforms like Coursera, edX, and Udemy — has grown to a $22.8 billion market, projected to exceed $119 billion by 2029.

But the more interesting growth story is not in the mega-platforms. It is in the creator economy. Goldman Sachs and Precedence Research estimate the creator economy at $250 billion in 2025, growing toward $480 billion by 2027. The share of that economy generated by educational content (online courses, cohort programs, workshops, and certifications) is growing faster than the entertainment segment it once trailed.

The Kajabi 2025 State of Creator Commerce report² captures a cultural shift the numbers alone don't show. The creator who monetizes through sponsorships and ad revenue is giving way to the creator who sells expertise directly, on their own terms.

Cobb's book publishing analogy is exactly right, and I've watched the same arc play out on Ruzuku. Self-publishing exploded when Amazon made distribution trivially cheap. For a few years, being first gave independent authors an advantage. Then the market matured, reader expectations rose, and the advantage returned to creators with genuine audiences and demonstrated expertise. Online courses followed the same arc, about five years later. In The Business of Courses, I call this the shift from "information scarcity" to "attention scarcity." It's the single most important market dynamic course creators need to understand.

The scale numbers tell a story. Coursera has surpassed 175 million registered learners. Udemy has 69 million students. Kajabi reports 60 million across its creator ecosystem. Ruzuku's own data shows 853,572 unique students enrolled across 32,000+ published courses. We're smaller than the mega-platforms by design, which reflects the kind of creators we serve: serious professionals, not mass-market content producers.

The workforce dimension adds urgency. Ninety percent of companies now offer some form of eLearning to employees. The demand for online learning is structural, not cyclical. This is what gives me long-term confidence in this market, even as the easy-money narrative fades: the underlying need for reskilling is accelerating, not slowing down.

So the market is massive. But I'm less interested in the macro numbers than in what individual creators actually charge and earn. That's where the data gets interesting.

Chapter 2: What Do Online Courses Actually Cost?

Course pricing data from major platforms is rarely published in detail, because it would reveal uncomfortable things about how most creators price, and how most students respond. Ruzuku's production data offers an unusually complete picture: pricing information from 175,248 price options across 32,000+ published courses.

What I find most interesting about the pricing distribution is the gap between median and mean. That nearly 4x spread is not a statistical anomaly. It's structural. Coaching programs priced at $2,000-$5,000 exist in the same database as yoga sequences priced at $47. "Online course" is not a single product category. It's a container that holds everything from impulse purchases to major professional investments.

The interquartile range (the middle 50% of paid courses) falls between $50 and $333. This is the "typical" course price range for most creators. Within this range, the median one-time payment is $99, payment plan median is $212, and subscription median is $49.99 per month.

Niche matters enormously for pricing. This is something I emphasize in The Business of Courses: your pricing ceiling is set by your niche and the specificity of the transformation you promise, not by how much content you include.

Look at the gap between coaching ($531 median) and writing ($70 median). That's a 7.6x difference, and it has nothing to do with content volume. Coaching programs promise high-stakes transformation: career change, business growth, personal breakthroughs. Writing courses compete with free YouTube tutorials and $15 Udemy offerings. Niche specificity is the pricing lever. A coaching program with 8 modules and weekly live sessions can price at $2,000-$5,000. A yoga course with 40 recorded classes is unlikely to break $300.

O'Lochlainn's point, prioritize depth over breadth in your client relationships, is supported by the platform data. Creators on Ruzuku have a median of 8 published courses. The model that generates sustainable revenue is not "launch one course and sell it forever." It is building a curriculum: a progression of offerings that take students from entry-level to advanced, from one-time purchases to ongoing relationships. I've seen this pattern play out consistently: the creators who build sustainable businesses are the ones who think in terms of a curriculum, not a single product. Dorie Clark built a $2 million revenue stream from video courses filmed on her iPhone, but that success was built on years of audience development and a multi-product curriculum.

For a deeper dive into pricing strategy, see The Complete Guide to Course Pricing or use the Revenue Calculator to model your own scenarios.

Pricing tells you what creators charge. It doesn't tell you whether students actually get value. And that brings us to the most uncomfortable finding in the data.

Chapter 3: Why Do Most Students Never Finish?

I have a line I use with creators: enrollment tells you how good your marketing is. Completion tells you how good your course is. By that measure, most courses are not very good. The numbers are worse than most people in the industry publicly admit. Class Central's analysis³ of massive open online courses found a median completion rate of 12.6%. For self-paced courses more broadly, the range is 10–20%. These numbers have been stable for years — not because the platforms haven't tried to improve them, but because they've been attacking the wrong variable.

Compare that to cohort-based programs. Dr. Philippa Hardman⁸, a learning design researcher at Cambridge, has documented completion rates of 96% in altMBA cohorts and 70%+ in Section4 programs — both structured around active cohort participation rather than passive content consumption. The difference isn't content quality. It's social architecture.

The conventional response to poor completion has been content optimization: shorter videos, more quizzes, gamification points, drip schedules, reminder emails. These interventions are not wrong. They are just insufficient. They treat a social problem as a content problem. I've watched this play out thousands of times on our platform — creators spending months perfecting their video quality, agonizing over lesson length, building elaborate drip sequences. And their completion rates barely move.

In 14 years of watching courses on our platform, what surprised even me was the magnitude of this effect. We expected community to help. We did not expect it to be the single most predictive structural element — more predictive than course length, production quality, price point, or instructor credentials. That finding reframed my entire understanding of what online courses actually are. They are not content delivery systems. They are social structures that happen to include content.

The research backs this up. Garrison, Anderson, and Archer's Community of Inquiry framework¹³, the foundational model for effective online learning, identifies exactly three presences that make online education work: teaching presence, social presence, and cognitive presence. Social presence is not incidental. It is one of the three pillars.

The effect compounds when you add cohort structure. Scheduled courses, where students move through material together on a defined timeline, see 64.2% completion compared to 48.2% for open-access courses. The combination of community discussion and cohort scheduling creates the most powerful completion engine available to independent course creators. A meta-analysis by Richardson, Maeda, Lv, and Caskurlu (2017)¹⁴ found social presence positively associated with both student satisfaction and perceived learning outcomes across 49 studies — independent validation of why the community effect in our platform data is as robust as it is.

Here's another data point: 43.2% of published courses on the platform have at least one student comment. Those courses with active discussions average 158.8 comments per course. Once engagement takes hold, it takes hold in a big way. The single most commented-on course on the platform has 185,151 comments. For the courses where community works, it is the core of the experience.

Shapiro's 90% completion rate in Truce with Food is what happens when you design community into the course from the beginning, creating discussion spaces where students process their experiences together, not in isolation. Her insight about "adaptive challenges" versus "technical solutions" maps onto the platform data exactly. This is the core of what I mean by transformation over information. If you're only transferring information, completion doesn't matter much. If you're facilitating genuine change, completion is everything.

I wrote a full analysis of completion rate factors and how to close the gap in your own courses: The Completion Gap: Why Most Students Don't Finish (And What to Do About It).

Community is the strongest predictor of completion. That raises a bigger question: what if community isn't just a feature? What if it's the product?

Chapter 4: Why Is Community Becoming the Product?

Ten years ago, community was a feature: a forum bolted onto a course to give students somewhere to ask questions. Today, the platform data and the Kajabi reports both point the same direction: community is the product itself, and course content is the delivery mechanism. I've watched this shift happen gradually on Ruzuku, and then all at once.

The Kajabi 2025 report² found that creators who include community elements earn 2x more than those who do not. The correlation is isn't causation, but the pattern is consistent across every data source I've seen: the shift from "sell content" to "sell transformation through community" is the clearest differentiator between creators who thrive and those who stall.

Ruzuku's own data puts a number on this: 9,140,636 total comments across all published courses, against 19,842,975 lesson completions. That's roughly one comment for every two lesson completions. Students are not just watching and clicking "complete." They're processing, reflecting, and responding.

Bianchini built Mighty Networks around the thesis that community is more durable than content, so she has skin in this game. But the platform data confirms her prediction: community and courses are converging. I'd go further. Community was never separate from learning. The industrial model of education just made us forget that. This is what I mean by transformation over information: the transformation happens between people, not between a person and a screen.

The customer stories from Ruzuku's support data reinforce this at the individual level. Barrie Risman, a yoga instructor, chose Ruzuku explicitly because she wanted community features to differentiate her program. "I love the community emphasis — this is something I want to set me apart," she told the support team. Andrea Thibaudeau put it even more directly: "This sense of community is EXACTLY why I switched to Ruzuku." These are not passive consumers of a feature. They are creators who have made community the center of their business strategy.

This has real implications for platform choice. A platform that treats discussions as an add-on (a separate forum, a Discord server linked from a welcome email) is not the same as a platform where discussion is woven into the learning flow. See how Ruzuku's community features are structured around the learning, not beside it.

Keck is right, and the research backs him up. In my HCI work at UNC, we studied how people's behavior changes when their actions are visible to peers. The same principle applies in a course discussion: when students know others can see their work, they show up differently. Researchers call this social translucence, and it's one of the most reliable drivers of sustained engagement. A 2022 study applying Self-Determination Theory to online learning¹⁶ found relatedness to be the strongest predictor of engagement across behavioral, emotional, and agentic dimensions (β=.52–.55 across 1,201 students) — the psychological mechanism behind what Keck observes in practice.

O'Lochlainn identifies something beyond completion: loyalty. A student who has built relationships with peers in your program, who is known by name in your discussions, who has invested emotionally in the group? That student does not leave when a competitor offers a slightly lower price. The economic moat of community is harder to build than better content, and far harder to replicate. In The Business of Courses, I call this the Retention stage of the Customer Learning Journey. It's where the real economics of course businesses live.

The research on belonging in online education reinforces this. Peacock et al. (2020)¹⁵ found that sense of belonging is directly linked to attainment, satisfaction, and persistence in online programs. Missing discussions were one of the primary factors that weakened it. Students who feel they belong to a learning community are not just more likely to complete. They are more likely to return, to refer others, and to purchase the next program. Belonging is not a soft metric. It is the foundation of the retention economics that sustain a course business.

Community is the most durable competitive advantage a course creator can build. AI is the most disruptive force hitting the industry. The interesting part is how these two forces interact.

Chapter 5: How Is AI Changing Course Creation?

Here's what I think most AI commentary gets wrong about course creation: the debate is framed as replacement versus irrelevance. AI will replace human instructors entirely, or AI is just a fancy autocomplete that serious creators should ignore. Both positions are wrong. The actual picture is more interesting, and more useful, than either extreme.

The efficiency gains are real and significant. Creators report 40–60% reductions in production time for standard content creation tasks — research, first drafts, social copy, email sequences. Tasks that once took half a day take two hours. For a solo creator who is simultaneously subject matter expert, content producer, marketer, and customer service department, that time savings is genuinely significant.

Cobb nails the distinction. AI can produce competent explanations of almost any topic. What it cannot produce is the credibility that comes from having done the work: having coached 1,000 clients through a specific transformation, having failed publicly and learned from it, having a track record students can verify. This is the information-versus-transformation distinction again. AI commoditizes information. Transformation still requires a human on the other end.

Dr. Philippa Hardman⁸, a learning design researcher at Cambridge, has studied AI's impact on the instructional design profession extensively. Her conclusion: AI is highly effective at automating the retrieval and explanation of established knowledge, but cannot replicate what she calls adaptive instructional presence: the real-time judgment calls that experienced instructors make when students are struggling. What should I say to this specific student, right now, given everything I know about how they learn? That's still a human function. Put simply: AI is excellent at retrieval and explanation. Humans remain essential for judgment and connection.

Here's how I'd frame the practical question: use AI for the production tasks where human judgment adds little value (first drafts, transcription, formatting, email sequences). Then invest the time you save into the human elements AI cannot replicate: live sessions, discussion responses, personalized feedback, community presence. The creators who will struggle are the ones who use AI to automate the human parts and leave only content behind. The ones who will thrive use AI to accelerate production so they can afford to be more present, not less.

The reskilling numbers make this even more urgent. With more than 50% of the global workforce needing significant upskilling by 2026, and 81% of career changers crediting online courses with enabling their transition (FutureLearn), the demand for human-led, expert-guided learning is growing faster than AI can commoditize generic educational content. The niche expert with deep domain knowledge and a track record of student outcomes is not threatened by AI. The generic information broker is.

Those are the macro forces. Now let me show you what the best creators actually do with them.

Chapter 6: What Do the Best Courses Have in Common?

Across 32,000+ published courses, the platform data and Course Lab interviews point to five patterns that consistently distinguish the courses with the best outcomes, for students (completion, transformation) and creators (revenue, retention, referrals) alike. At this point, I can predict with reasonable confidence whether a new course will succeed based on how many of these patterns it incorporates.

Pattern 1: They design for transformation, not information

The most common failure mode for new course creators is building a course that covers everything they know about a topic, organized by subject matter the way a textbook would be. This is an information architecture, not a learning design. Students don't need more information. They need to change their behavior, build a skill, or achieve a specific outcome.

Stever Robbins, executive coach and founder of Get-it-Done Guy, calls this the"knowing-doing gap." Most students who enroll in a course already know, at some level, what they need to do. They are paying for structured support in actually doing it. This is the core argument of The Business of Courses: if it was just about the information, you wouldn't need the course. YouTube has the information. What students are paying for is the change.

Jonathan Levy, author of The Only Skill That Matters, frames it from the student's perspective: "What people are paying for is the curated experience. They can find almost any information for free. They can't find someone to organize it, sequence it, and hold them accountable through it."

The learning science here is unambiguous. Chi and Wylie's ICAP framework (2014)¹⁷ establishes a hierarchy of cognitive engagement: Interactive > Constructive > Active > Passive. Passive reception of information (watching videos, reading slides) produces the weakest learning outcomes. Constructive activities (applying concepts, generating responses) and interactive activities (discussing, debating, teaching others) produce the strongest. Courses that rely entirely on video consumption are asking students to learn in the least effective mode available. Courses that build in discussion, application, and peer feedback are structurally aligned with how learning actually works.

Pattern 2: They price for the outcome, not the content

The platform pricing data supports what experienced creators have learned empirically: students pay for results, not for hours of video. A yoga course that promises "200 hours of recorded practice" is competing on a commodity dimension. A yoga course that promises "a sustainable daily practice in 30 days, or your money back" is competing on a transformation dimension, and can price accordingly.

The coaching niche's $531 median price reflects this principle. Coaching courses are not 5x more valuable because they have 5x more content. They command a premium because the promised transformation (career change, business launch, personal breakthrough) is perceived as 5x more important to the buyer. See the Ruzuku for Coaches page for examples of how coaching programs are structured on the platform.

Pattern 3: They build community into the curriculum

This pattern underlies everything discussed in the completion rate chapter above. The courses with the best outcomes treat community discussion not as a forum where students can optionally post questions, but as a structural element of the learning design. Discussion prompts are built into modules. Peer feedback is a required step. The instructor's presence in discussions is predictable and responsive.

Jerry Fresia's fine arts courses on Ruzuku illustrate this at its most intentional. His entire pedagogy is built around critique: students post their in-progress paintings in discussion threads; other students and Jerry himself respond with feedback. The discussion is not supplementary to the learning — it is the primary mechanism through which learning happens. I've watched Jerry's courses for years, and the completion pattern is striking: students who post their first painting almost always finish the course. The 185,151-comment course mentioned earlier? That's Jerry.

Pattern 4: They create re-entry points

Students drop out. This is a fact, not a failure. The question is whether your course design makes it easy or hard to come back. Courses that treat absence as abandonment — that have no mechanism for students who fall behind to rejoin without embarrassment, those have much higher permanent dropout rates than courses that normalize returning.

Kamb's "respawning" concept captures something I've watched play out hundreds of times on Ruzuku. In a self-paced course without community, missing a day feels like falling behind. Missing a week feels like failure. Many students never come back. But in a community, missing a session is normal. Others have missed sessions too, and the group welcomes them back. This is what I call the re-entry dimension of the Customer Learning Journey: retention is not about preventing dropout. It's about making it easy to come back. Instructors who build explicit re-entry mechanisms into their courses (mid-course reset weeks, "come back" email sequences, cohort models where new starts happen regularly) retain students who would otherwise quietly disappear.

Pattern 5: They measure outcomes, not enrollments

Enrollment tells you how good your marketing is. Completion tells you how good your course is. The creators with the best long-term businesses are obsessive about the second number, not the first. A course with 200 completions and strong testimonials is more valuable (in word-of-mouth, referrals, and pricing power) than a course with 2,000 enrollments and 12% completion. Ali Shapiro's 90% completion rate isn't just a statistic. It's the reason her courses command premium prices and consistently sell out. The transformation she delivers is verifiable.

You can see these patterns across creator case studies on the platform. Michael Sheridan grew his dream interpretation course business from $125K to $1.2M by adding live group coaching to recorded modules (patterns 1, 3, and 5 simultaneously). Amy Medling at PCOS Diva built recurring seasonal programs with dedicated community support (patterns 3 and 4 in combination). These are not outliers. They are the predictable result of designing for transformation, not information.

The patterns are clear. Here's what I'd do with them.

What This Means for Course Creators in 2026

The industry is not dying. It's maturing. The conditions that made early success easy (information scarcity, low competition, novelty as a selling point) have given way to conditions that favor expertise, trust, and genuine student outcomes. If you built a business on those early advantages and didn't adapt, the "gold rush is over" narrative feels true. If you're building around community, specific transformation, and measurable results, you're experiencing something different.

The Ruzuku platform data supports this: $86 million in total revenue across 853,000 students is real economic activity, not hype. These are independent creators teaching what they know and serving students who actually finish.

If I were starting a course business today, here's what I'd do: pick a niche where I have genuine expertise and earned credibility. Build a community-driven program around a specific transformation. Use AI to accelerate production so I can invest more time in live interaction, not less. Measure success by student outcomes, not enrollment numbers. Aim for 100 true fans who complete and refer, not 10,000 leads who bounce.

That model is harder to build than a library of recorded videos. It's also far more durable, more profitable, and more meaningful. That's the thesis of The Business of Courses, and after 14 years and 32,000 courses, I'm more convinced of it than ever.

Methodology

The Ruzuku platform data cited in this report covers the full production database as of early 2026: 32,000+ published courses, 74,840 total courses (including drafts), 853,572 unique students, 1,826,838 total enrollments, 9,140,636 total comments, and $78,866,538 in total platform revenue. All pricing data is in USD unless otherwise noted; the platform also processes payments in GBP, CAD, AUD, and EUR.

Completion rate figures compare the percentage of enrolled students who completed all required steps within a course, segmented by whether the course had active community discussion (at least one student comment) and whether it was structured as a scheduled cohort versus open-access self-paced. These segmentations are observational, not experimental — they reflect associations in the data, not randomized controlled trials.

Limitations. Ruzuku's creator base skews toward professional service providers (coaching, health and wellness, creative arts, professional development) rather than the mass-market consumer courses that dominate larger platforms like Coursera and Udemy. Platform data should not be extrapolated directly to the full market. Self-selection bias is present in all completion rate comparisons: creators who choose to build community discussion into their courses may also differ from those who don't in other ways that affect completion (more engaged instructors, higher-touch formats, more motivated student cohorts).

External statistics are sourced to named reports and research organizations. Market size figures from Statista, HolonIQ, IMARC Group, and Mordor Intelligence use different methodological scopes (total eLearning vs. corporate eLearning vs. MOOC segment) and should not be added together. Creator economy figures from Goldman Sachs and Precedence Research include broader content categories beyond educational courses.

This report was last updated March 2026. Statistics will be updated annually.