How to structure an online course: modules, lessons, length
How to split a video course into modules and short lessons: a study of 6.9 million video sessions (edX) points to chunks under 6 minutes. A template to copy.
A good structure is half of a good course: it decides what students see, in which order, and how hard they work to reach the end. This guide proposes a simple method for splitting the material, with the little evidence we read and its limits.
An honest premise: the most-cited data on video length come from an observational study of 2012 edX science courses at university level. We found no equivalent study on professional courses sold by creators.
Where do you start to structure a course?
You start from the result: what the student can do at the end that they could not before. Then you list the skills needed to get there and group them into modules. The structure follows the result, not the order in which ideas occurred to you.
Write the result in one checkable sentence (“by the end you will be able to publish a site on your own domain”) and list the steps needed. Each module should have a goal the student can check alone, for example a small piece of work to hand in. If a module does not lead to the course goal, remove it or move it to extra materials. For the whole path, from recording to selling, read how to create and sell an online course.
How long should lessons be?
Guo, Kim and Rubin's study (2014) analysed 862 videos from four edX courses, with 127,839 students and about 6.9 million viewing sessions. Video length turned out to be the strongest predictor of engagement: median watching time never exceeds 6 minutes, whatever the video length.
Share who attempt the problem after the video, by length group
| Length group | Percentage |
|---|---|
| Shortest videos (first group) | 56% |
| Second group | 48% |
| Third group | 43% |
| Fourth group | 41% |
| Longest videos (fifth group) | 31% |
The authors recommend segmenting videos into short chunks, ideally under 6 minutes, and admit that shorter videos may also be better planned: they lack the data to separate the two effects. It is a retrospective study, not a controlled experiment, limited to 2012 maths and science courses with probably highly motivated students. Use it as a hint for theory lessons, and check with your own students by looking at where the viewing graph drops.
Which lesson format works best?
The same study suggests that an instructor appearing between slides engages more than slides alone, that a personal tone can beat studio production and that hand drawing on a tablet beats static presentations. These are observations, with only one comparison between two courses.
The authors also note that faster, more enthusiastic speakers engage more: speaking rate ranged from 48 to 254 words per minute, with a mean of 156, and within a length group engagement rose up to double with speaking rate. That is no reason to rush: it means a lesson read in a monotone loses attention. A practical consequence is to rehearse the lesson aloud before recording and to cut pauses and repetitions that serve no purpose. Audio quality and clarity matter more than a TV studio.
How do you balance videos, exercises and materials?
You alternate: a short lesson, a short exercise, a check at the end of the module. In the edX study 32% of videos were immediately followed by a problem to solve, and the share who attempted it fell with video length.
A typical module has five to seven lessons, one practical exercise, a one-page written recap and a small assignment. Keep materials (worksheets, templates, files) in a single place in the student area, not inside videos. For practical courses the study notes viewers watch on average only 2-3 minutes of each tutorial, whatever its length, and rewatch more often: so clear chapters, descriptive titles and the ability to jump to the step of interest are needed. For video hosting read where to host online course videos.
How many people reach the end of a course?
Few, in open courses. Jordan's (2015) analysis of 221 massive open online courses (MOOCs) found a median completion of 12.6% of enrolments, with a range from 0.7% to 52.1%, but with differing definitions of “completion” and data from 2013 or earlier.
The author observes that longer courses had lower completion and infers, as her own inference rather than a tested result, that a case could be made for shorter, more modular courses. The figure does not apply to a creator's paid course: people who pay have different motivations from those who sign up for a free course. For your own course the measure to watch is the percentage of students who finish each module, and where they stop. If the drop is at a specific point, rework that lesson. For the cohort versus self-paced comparison read cohort or self-paced course.
How do you build a structure to copy?
A simple template: an introductory start module with a quick small win, four to six core modules each with a goal, a final module with a project and a plan for continuing. Each module contains short lessons, an exercise and a recap.
The start module matters more than it seems: in Jordan's data the first two weeks are the critical moment when most enrolled people drop away. Give students a first result in a few minutes, for example a small file handed in, so they see the course works. Write the syllabus on the sales page with the same modules and goals, and leave out what does not serve the result. For course pricing see how much to charge for an online course.
Key terms. A module is a group of lessons with a common goal. A lesson is a short unit of content. Engagement time is the length of a viewing session. Completion rate is the share of enrolled people who finish the course.
How we checked this. We read the full Guo, Kim and Rubin (2014) study and Jordan's (2015) analysis and drew the numbers above from them, with the authors' caveats. We could not read as text the figure with the exact limits of the length groups and found no equivalent studies on professional creator courses, and we say so. Contact: hi@heytea.app. Editorial note: we fact-check every figure against the sources listed below and correct mistakes openly.
Frequently asked questions
How long should a video lesson be?
The most-cited study, on 6.9 million sessions of edX videos, found median watching time never exceeds 6 minutes, whatever the video length, and recommends segments under 6 minutes. It is an observational study of 2012 university science courses: it shows a direction, not a rule for every subject.
How many modules should a course have?
There is no right number. Start from the result the student must achieve, list the skills needed to get there and group them into 4-8 modules, each with a checkable goal. If a module has more than eight lessons or more than one goal, split it.
Better to record lessons in a studio or informally?
In the same edX study, videos with a more personal tone, recorded at a desk, had more engagement than those shot in a TV studio, especially beyond 6 minutes. The authors had only one comparison of two courses, though: it is a hint, not proof. Clarity matters more than production.
Do you need exercises after each lesson?
They help. In the edX study 32% of videos were immediately followed by a problem to solve, and the share of people attempting it fell with video length, from 56% for the shortest to 31% for the longest. A short exercise after each lesson gives students a way to check what they understood.
How do you adapt the structure to a practical course?
A practical course works better as a sequence of tutorials: each lesson produces a visible result. The study notes viewers watch on average only 2-3 minutes of each tutorial and rewatch more often, so clear chapters and a way to jump to the step of interest are needed.
Sources
- Guo, Kim, Rubin (2014), “How Video Production Affects Student Engagement: An Empirical Study of MOOC Videos”, ACM Learning at Scale. Read on 16 October 2026. up.csail.mit.edu
- Jordan (2015), “Massive Open Online Course Completion Rates Revisited”, IRRODL 16(3). Read on 16 October 2026. irrodl.org