How does recorded completion differ across course formats on Ruzuku? This analysis compares existing Ruzuku course groups. It does not establish a completion advantage over marketplaces or university MOOCs, whose populations and completion definitions differ.
To answer this question, we analyzed a single snapshot of Ruzuku’s production database (copied 20 August 2026, analyzed 1 September 2026) covering 27,156 published courses, 1,316,272 student enrollments, 14,185,137 lesson completions, and 5,111,472 discussion comments. It is a snapshot, not a longitudinal study, so it does not show trends over time.
This paper compares recorded completion across five things: delivery format (scheduled, on-demand, or open-access), the number of students in a scheduled course, whether lessons have discussions, whether the course has assessments, and whether a student completed a lesson in the first week.
These observational records can inform learning-design questions, but they do not validate a design framework or prove why students complete a course. Our teaching recommendations draw on practitioner experience as well as these records; the dataset does not isolate those recommendations’ effects.
Recorded completion rate across 337,700 scheduled cohort enrollments.
Ruzuku platform data
Highest of five size bands. Averaged across scheduled courses, so not comparable with the 53.4% figure.
2,716 Analyzed Courses
August 2026 snapshot: 50.9% recorded completion per enrollment with an enabled lesson discussion; 37.3% without. Observational, not causal.
17,190 courses with an enabled lesson discussion
78.2% completion when a lesson was completed in the first 7 days, 17.2% when none was. Observational.
921,131 enrollments
Executive Summary: Core Statistical Findings
Key Findings from 1.3 Million Enrollments
- 1.Course Format Groups: Scheduled cohort courses achieve a 53.4% average completion rate, compared to 41.9% for open-access self-paced courses and 42.9% for on-demand drip programs.
- 2.Cohort Size Groups: Completion follows an inverted-U curve. Cohorts with 11–25 learners peak at 65.2% completion (2,716 courses), with higher recorded completion than smaller groups of 1–10 (51.5%) and larger groups of 100+ (51.3%).
- 3.Discussion and Assessment Groups: Courses with an enabled lesson discussion had 50.9% recorded completion per enrollment, versus 37.3% without. Courses with assessments had 59.1%, versus 42.4% without. These August 2026 Ruzuku comparisons are observational; they do not isolate the effect of adding either feature.
- 4.First-Week Activity: Enrollments where the student completed at least one lesson in the first 7 days recorded 78.2% completion, compared with 17.2% where no lesson was completed that week (a 4.55x difference). This is an association, not a cause.
- 5.Niche Differences: The format pattern is not the same in every subject area. Scheduled courses recorded higher completion than open-access courses in some niches, such as coaching, and lower completion in others, such as therapy and yoga. Niche samples are much smaller than the platform totals.
The Teaching Frameworks Behind My Recommendations
Why do students drop out of some online courses while finishing others? In our World-Class Courses workshops, we teach that online learners do not abandon courses due to lack of intelligence or discipline. They drop out because of predictable cognitive traps and flawed instructional architecture:
1. The Tyranny of Hyperbolic Discounting & Temporal Scaffolding
Behavioral economics demonstrates that humans systematically overvalue immediate comfort over delayed long-term rewards (hyperbolic discounting and present bias). In an open-ended self-paced course with lifetime access, there is never an urgent reason to study today. A student tells themselves, “I will finish Module 3 on Saturday,” but Saturday brings immediate competing demands. Scheduled cohort milestones provide the temporal scaffolding that makes present action salient and overcomes procrastination.
2. The Two Traps of Ineffective Learning Design
Trap 1: The Instructionism Trap assumes that teaching is “telling.” Driven by the Curse of Knowledge (experts forgetting what it is like to be a novice), creators record 45-minute monologue lectures. Trap 2: The Aboutness Trap conceives of a course as being “about” a subject taxonomy rather than guiding a student through a personal progression. This leads to the opposite of what author Jen Louden calls “appropriate dosing”—overloading students with too much content, too fast, triggering cognitive overload by Module 2.
3. Dual Minimalism: The Antidote to Course Bloat
The structural antidote to both traps is Outcome-Focused Design with Dual Minimalism: (1) defining tightly focused, well-scoped outcomes that deliver tangible real-world value, and (2) providing the minimal viable content necessary to scaffold progress toward those outcomes. Using our “Two-Hour Plane Ride Question”, creators strip away encyclopedic filler and focus exclusively on core levers of transformation.
4. The Community of Inquiry (CoI) & Social Translucence
Grounded in Garrison, Anderson, and Archer’s Community of Inquiry (CoI) framework, deep online learning requires three interlocking presences: Teaching Presence (expert facilitation and curriculum structure), Cognitive Presence (critical reflection and concept articulation via the generation effect), and Social Presence (the ability of participants to project themselves as real people). In-lesson discussion feeds create social translucence—making peer progress visible and eliminating the isolation of solo digital study.
A Minimum Viable Offer & The 4 Scoping Questions
Before an educator records a single video or designs a curriculum, high completion rates must be engineered into the course offer. A useful way to think about this is a minimum viable offer: before you build, resolve the two primary failure points of learning design:
- 1. Creating a course “about” a topic vs. solving a specific problem or improving performance, results, or experiences. When you define a course by topic taxonomy, you accumulate endless informational content. When you define it by a concrete capability, you can cut the lectures that don’t serve that capability.
- 2. “Overdosing” vs. appropriate dosing (Jen Louden). Course creators frequently suffer from the Curse of Knowledge, overwhelming students with more theory than human working memory can process in a week.
To solve these problems, start by asking yourself: based on everything you have learned from your research and conversations with prospective students, what do people really want from your course? What is the solution they need, or the better experience they seek?
One creative technique is “Magic Wanding”—imagining the ideal, unconstrained outcome a student desires. But you must ground that vision in real research calls, and ask what you can realistically help students achieve in the scope of your pilot.
The 4 Scoping Questions That Refine a High-Completion Pilot
Question 1: The Duration Boundary
How long will my pilot have to be in order to deliver on this promise?
There are exceptions, but in most cases pilots should not be more than 4 to 6 weeks long. If you need much longer to deliver on your promise, scale back what you are thinking about delivering to maintain student urgency and momentum.
Question 2: The 30-Day Early Quit Test
If someone were to start my pilot but quit after a month, would they see any benefits? If so, what would those be? And how could you cram the most possible benefits into those first four weeks? Can you make those the only benefits offered in the pilot?
Question 3: The First Visceral Milestone
If whatever you are doing for them is yielding results, what would be the first big milestone they would hit that gives them a real visceral sense that this is working? If that milestone is so compelling, can you make it the entire promise of your pilot?
Question 4: The Two-Hour Plane Ride Test
If you were on a two-hour plane ride with your ideal customer, what could you teach them in those two hours that would leave them saying, ‘Wow, that really helped me’? Can you limit the scope of your pilot to whatever you would teach in those two hours?
1. Delivery Format Benchmarks: Scheduled Cohort vs. Self-Paced
In instructional systems design, delivery models are categorized into student-directed asynchronous (self-paced) and calendar-directed synchronous/hybrid (cohort-based). On Ruzuku, each course records its access format as a course setting:
| Delivery Format | Total Analyzed Enrollments | Empirical Completion Rate | Observed Difference vs Open Access | Format Characteristics |
|---|---|---|---|---|
| Scheduled Cohort Courses Calendar-based course access; live delivery is not classified | 337,700 | 53.4% | 11.5 percentage points (observed) | Scheduled calendar; live sessions not isolated |
| On-Demand Drip Courses Automated interval releases from enrollment date | 213,519 | 42.9% | 1.0 percentage point (observed) | Interval releases from enrollment date |
| Open-Access Self-Paced Immediate full content access, self-directed timeline | 702,497 | 41.9% | Baseline (1.0x) | Self-directed schedule; discussion presence varies |
In the August 2026 Ruzuku snapshot, scheduled courses had 53.4% recorded completion per enrollment, compared with 42.9% for on-demand and 41.9% for open-access courses. These are observational course-group differences, not evidence that deadlines or live sessions caused higher completion. The data does not compare vendors.
2. Observed Completion Across Cohort Sizes
A common assumption in course creation is that smaller cohorts always yield higher completion because facilitators can offer more personalized attention. Our data reveals that this assumption is incomplete.
Grouping 9,559 scheduled courses into five enrollment bands shows an inverted-U curve. Each band’s rate is the average across its courses:
| Cohort Size Band | Course Sample Size | Empirical Completion Rate | Design Considerations (Not Tested Explanations of Completion) |
|---|---|---|---|
| 1 – 10 Students | 4,052 courses | 51.5% | High individual attention. Discussion can stall with only a few active commenters, so plan prompts that work for a small group. |
| 11 – 25 Students | 2,716 courses | 65.2% | Consider the amount of discussion and feedback your facilitator can support; this group’s completion rate does not establish optimal peer interaction or psychological safety. |
| 26 – 50 Students | 1,400 courses | 62.0% | Strong community momentum; requires active moderator facilitation to maintain thread organization and prevent lurker disengagement. |
| 51 – 100 Students | 712 courses | 56.8% | Discussion volume accelerates; individual student visibility begins to erode; facilitator feedback shifts from individual to cohort-level. |
| 100+ Students | 679 courses | 51.3% | Plan how you will notice a student who stops showing up. At this size, neither peers nor the facilitator may notice. |
A Facilitation Consideration, Not a Tested Completion Mechanism:
In cohorts of 11 to 25 learners, a few students often become vocal “discussion drivers,” generating sufficient conversational energy that intermediate and quiet students feel part of a lively community. Because the group remains under 25, no single student feels anonymous, and the facilitator can realistically acknowledge every submission.
3. Discussions and Assessments
Beyond scheduling and class size, the dataset also compares recorded completion in course groups with and without discussions or assessments:
50.9% vs. 37.3% Completion (+13.6 pts)
In the August 2026 Ruzuku snapshot, courses with at least one enabled lesson discussion had 50.9% recorded completion per enrollment, compared with 37.3% for courses without. This observational comparison does not establish cause or compare vendors. The groups comprised 17,190 courses with an enabled lesson discussion and 9,442 without. Enabled does not mean students actively participated. See the Course Success Index.
This comparison sorts courses only by whether a lesson discussion was turned on. It does not use the comments themselves. In my experience, when students can see peers reflecting on the same lesson, studying online feels less solitary.
59.1% vs. 42.4% Completion (+16.7 pts)
Across 26,632 courses, courses with at least one assessment lesson recorded 59.1% completion (5,598 courses), compared with 42.4% for courses without assessments (21,034 courses).
Formative knowledge checks trigger the testing effect, reinforcing memory retention, proving competence, and providing clear psychological closure before unlocking subsequent modules.
4. First-Week Activity and Completion: A 4.55x Difference
When does a student decide to finish—or abandon—an online course?
I compared 921,131 enrollments, in courses that had new students in the five years before the analysis, by whether the student completed a lesson within 7 days of enrolling:
| First 7 Days After Enrolling | Enrollments | Recorded Completion | Difference |
|---|---|---|---|
| Active in Week 1 (≥1 Lesson Completed in Days 0–7) | 352,713 | 78.2% | +61.0 percentage points |
| Inactive in Week 1 (0 Lessons Completed in Days 0–7) | 568,418 | 17.2% | Baseline |
Instructional Implication: Watch the First Week
Students who completed at least one lesson within their first week recorded a completion rate 4.55 times higher than students who did not. That is an observed association, not proof that the first lesson caused the difference. Part of the gap is built in. Completing a lesson is already progress toward completing the course, and in a one-lesson course it is the same thing. In World-Class Courses, we instruct creators to apply the “Know-Feel-Do” onboarding rule: ensure Module 1 produces an immediate small win that creates emotional confidence (Feel) and verifiable action (Do) before introducing heavy theory (Know).
5. Niche by Niche: The Format Pattern Varies
The gap between scheduled and open-access courses is not the same in every subject area. In coaching, scheduled courses recorded higher completion than open-access courses. In therapy and yoga it ran the other way: open-access courses recorded higher completion than scheduled ones.
I assigned niches by keywords in course titles, so a course can land in the wrong niche, and many courses land in none. Niche groups are much smaller than the platform totals, and some hold only a few hundred enrollments in a given format. If you teach therapy or yoga, don’t use the platform-wide 53.4% figure as evidence that a cohort will help your students. Compare completion across your own runs instead.
6. The World-Class Course Design Rubric
In our World-Class Courses workshops, we provide creators with a design rubric. Use these five dimensions to review your current or upcoming online program. The rubric is teaching guidance; the completion data above does not score courses against it:
| Design Dimension | Flawed Architecture | Standard Online Course | World-Class Course |
|---|---|---|---|
| 1. Pedagogical Dosing & Scope | 60-minute video lectures; encyclopedic data dumps; passive slide readings (The Instructionism Trap). | 20-minute videos organized by topic chapters with downloadable PDF slide decks. | 5-to-12 minute micro-lessons chunked around a single concept paired with an immediate active application prompt (Dual Minimalism). |
| 2. Social Translucence | Zero community features, or discussion isolated in an external Discord/Facebook group that students never visit. | A general community forum tab disconnected from individual lesson pages. | Native point-of-learning discussion feeds anchored beneath every lesson, making peer reflection visible in real time. |
| 3. Temporal Scaffolding | Open-ended lifetime access with zero deadlines; students succumb to hyperbolic discounting. | Automated weekly email drip sequences without live cohort events or calendar checkpoints. | Scheduled cohort pacing with synchronized weekly milestones, calendar integration, and live video application labs. |
| 4. Formative Closure | Passive checkmarks without comprehension checks; students binge content without internalizing frameworks. | Optional end-of-course survey or high-stakes final exam with no mid-course feedback. | Frequent low-stakes self-check quizzes, reflection gates, and milestone credentials (The Testing Effect). |
| 5. Outcome Definition | Conceived as being “about” a topic (The Aboutness Trap). | Focuses on knowledge transfer (What students will Know). | Defined by what students will do, know, and feel (an approach Kelly Edmonds described on our podcast), aiming at a lasting change for the student. |
7. Downloadable CSV Dataset Summary
For academic researchers, instructional design consultants, and course creators performing custom meta-analyses, the platform-level figures from this study are summarized below:
12 rows × 7 columns · 1.1 KB · UTF-8 · Updated November 2026
Comparison,Group,Recorded Completion Rate,Rate Basis,Sample Size,Sample Unit,Snapshot Date Delivery format,Scheduled cohort,53.4%,per enrollment,337700,enrollments,2026-08-20 Delivery format,On-demand,42.9%,per enrollment,213519,enrollments,2026-08-20 Delivery format,Open-access self-paced,41.9%,per enrollment,702497,enrollments,2026-08-20 Lesson discussions,With an enabled lesson discussion,50.9%,per enrollment,17190,courses,2026-08-20 Lesson discussions,Without a lesson discussion,37.3%,per enrollment,9442,courses,2026-08-20 Assessments,With assessments,59.1%,per enrollment,5598,courses,2026-08-20 Assessments,Without assessments,42.4%,per enrollment,21034,courses,2026-08-20 Scheduled cohort size,1-10 students,51.5%,average of course rates,4052,courses,2026-08-20 Scheduled cohort size,11-25 students,65.2%,average of course rates,2716,courses,2026-08-20 Scheduled cohort size,26-50 students,62.0%,average of course rates,1400,courses,2026-08-20 Scheduled cohort size,51-100 students,56.8%,average of course rates,712,courses,2026-08-20 Scheduled cohort size,100+ students,51.3%,average of course rates,679,courses,2026-08-20
8. Four Instructional Design Options to Test
The observational findings and our practitioner experience suggest four design options to test. They do not predict a doubling of graduation or isolate the effect of any one change:
1. Keep Each Cohort to About 25 Students
The 11–25-student group had 65.2% recorded completion in this snapshot. That observation does not predict completion for a newly sized cohort. If you have 100 students, split them into four parallel 25-person cohorts.
2. Engineer an Unmissable First-Week “Quick Win”
The first-week activity groups had different recorded completion rates; this does not establish that activation caused the difference. As a practical onboarding step, ensure Module 1 can be completed in under 15 minutes and ends with an immediate interactive action prompt. Follow up automatically with students who have not logged in by Day 3.
3. Embed Point-of-Learning Discussion Prompts on Every Lesson
Never isolate community conversations into a detached external forum or social media group. Place discussion prompts beneath lesson videos so students can reflect and ask questions in context.
4. Integrate Formative Milestone Knowledge Checks
Use low-stakes self-check quizzes or reflection exercises to let students practice. The observed assessment-group completion gap does not establish what adding a quiz will do to your course’s completion.
9. The 5 Principles of Agile Course Development: From MVC to Responsive Teaching
High completion rates are not achieved through solitary perfectionism before launch. In our World-Class Courses mentorship, we train educators to adopt an agile, responsive development model:
1. Iterative Course Development (The Minimum Viable Course)
Never build a 40-hour curriculum upfront. Start with a “Minimum Viable Course” (MVC) released in agile weekly sprints. Gather real student reflection data from Module 1 before recording Module 3, allowing you to adapt future lessons to actual learner pacing.
2. Close Student Collaboration & Co-Creation
Treat early students as co-creators of the curriculum rather than passive consumers. Host regular live office hours and group feedback calls to uncover invisible cognitive hurdles. When students see their own questions answered in the next lesson, they tend to stay with the course.
3. Working Knowledge Over Perfect Content
Focus relentlessly on getting students actual, observable results first. In my experience, rough, usable exercises and short worksheets help students more than polished studio lectures. Start with core conceptual levers that create immediate value, and refine production polish over subsequent iterations.
4. The Responsive Teaching Model
Maintain the agility to adjust curriculum sequencing based on live cohort velocity. If several students stall on a specific assignment, record a 3-minute diagnostic loom or host a 15-minute live lab to dismantle the bottleneck in real time.
5. Student Progress Reviews & Cohort Retrospectives
Run a structured retrospective at the end of every cohort. Look at which lessons students actually completed and what they said in the discussions, rather than vanity metrics (like video views). Isolate the exact lesson where engagement dipped and streamline its cognitive load for the next cohort run.
10. Methodology & Data Provenance
Dataset Scope & Integrity Controls:
- • Source: One snapshot of Ruzuku’s production database, copied 20 August 2026 and analyzed 1 September 2026. It holds every enrollment on record at that date, old and new. It is a single snapshot, not a longitudinal panel, so it shows no trends over time.
- • Courses: 27,156 published courses with at least one lesson and at least one student. Deleted courses, templates, and bundles are excluded. No minimum size is applied, so a course with one student counts.
- • Enrollments: 1,316,272 student enrollments in those courses. Completion rates count only enrollments still enabled at the snapshot date (
is_enabled = true). That is why the three format groups add up to fewer enrollments than the total, and why the discussion and assessment groups cover 26,632 courses rather than 27,156. - • Completion: Binary platform-recorded completion (
course_completed_at IS NOT NULL). - • Weighting: Rates are per enrollment, so large courses count for more. Cohort-size bands are the exception: each course’s rate is computed first, then averaged across the courses in the band. Don’t compare a band’s rate with the 53.4% format figure.
- • Groups: Format is the course’s access setting: scheduled, on-demand, or open-access. Whether a course runs live sessions is not in the data. Cohort size is the number of enabled students in a scheduled course over its whole life, not the size of one run. “With a discussion” means at least one lesson has an enabled discussion, whether or not anyone posted. Niches are assigned by keywords in the course title only.
- • First week: Enrollments from any year in courses that had at least one new enrollment in the five years before the analysis. “Active” means the student completed a lesson within 7 days of enrolling. The clock starts at enrollment, not at the course start date, so a student who enrolls in a scheduled course before lessons open can count as inactive.
- • No controls: The comparisons do not adjust for price, course length, subject, or school. A gap between two groups can come from any of these.
- • Comments: The 5,111,472 comments are every comment in these courses’ discussions, by students and instructors alike. No comparison on this page uses them.
- • Disclosure: I co-founded Ruzuku and run it, and Ruzuku sells the features compared here. This analysis has not been peer reviewed.
11. How to Cite This Study
APA 7th Edition
Crystal, A. (2026). What 27,156 courses tell us about cohort vs. self-paced completion: An empirical analysis of 1.3 million online enrollments. Ruzuku. https://www.ruzuku.com/learn/articles/cohort-vs-self-paced-completion-study
BibTeX
@techreport{crystal2026cohortstudy,
author = {Crystal, Abe},
title = {What 27,156 Courses Tell Us About Cohort vs. Self-Paced Completion: An Empirical Analysis of 1.3 Million Online Enrollments},
institution = {Ruzuku},
year = {2026},
url = {https://www.ruzuku.com/learn/articles/cohort-vs-self-paced-completion-study}
}Run Your Next Cohort on Ruzuku
Ruzuku provides built-in browser video rooms, point-of-learning discussions, automated progress gating, and dedicated human student tech support—with zero platform transaction fees.
Related research: Explore our Course Completion Rate Benchmarks Guide, read about transformative design in Creating Transformative Online Courses, apply the 5 Keys to Student Engagement, or review student sentiment in our 2026 Course Platform Satisfaction Report.