How Long Should an Online Course Be? Sales vs Finishing

Longer courses sell better and finish worse. On Udemy, courses over 20 hours average 11,400 enrolments against 2,100 for courses under an hour. Meanwhile every extra week of course length costs about 1.75 percentage points of completion. Length is a sales asset and a delivery liability at the same time, and almost nobody writing about this admits it.

How long should an online course be?

For most coaching and course offers: two to four hours of finished video, split into modules of three to five lessons, with each lesson under six minutes where the material allows.

That answer is defensible but the reasoning matters more than the number, because the number changes depending on whether you are optimising for the sale or for the outcome.

What does the research on lesson length say?

One study dominates this entire subject. Guo, Kim and Rubin analysed 6.9 million video watching sessions across four edX courses and published at the ACM Learning at Scale conference in 2014.

Median engagement by video length, as summarised in Cindy Brame's peer reviewed review in CBE Life Sciences Education:

• Under 6 minutes: close to 100 percent of the video watched.
• 9 to 12 minutes: roughly 50 percent.
• 12 to 40 minutes: roughly 20 percent.
• Maximum median engagement time for a video of any length: about 6 minutes.

The other findings from the same study are less quoted and more actionable. Talking head interspersed with slides beat slides alone. Engagement rose with faster speaking rate, which the authors attributed to enthusiasm. Low production personal video beat studio production. And recorded classroom lectures failed even when segmented afterwards, which means planning short beats cutting short.

Be honest about what this is. It is 2014 data, from free STEM MOOCs, with university students, and it has never been replicated on paid creator courses. It is still the best evidence anyone has, which is why every article repeats the six minute figure. Most of them just do not tell you where it came from or how old it is.

Has anything challenged the six minute rule?

Yes, and it complicates the picture in a way that matters for course creators specifically. Wistia's 2026 State of Video report drew on more than 13 million videos and 79 million hours of viewing.

• Shorter videos do have higher engagement rates, consistent with Guo.
• But courses, educational video, webinars and podcasts hold attention better than other formats once you move into longer runtimes.
• Videos between 30 and 60 minutes drove the highest click through rate on interactive elements.
• Webinar replays of 31 to 45 minutes got more than twice the engagement of replays under 30 minutes.

Wistia's own framing is worth keeping: a 10 minute video with a lower engagement rate still delivers more total watch time than a 1 minute video with a higher one. Engagement rate is a ratio, and ratios can be gamed by making things shorter without teaching more.

This is B2B marketing video rather than paid course data, so it is a caution rather than a refutation. The safe reading is that six minutes is a good default for instructional lessons and a bad rule for everything else you might put in a course.

The statistics in this subject that are not real

Two claims turn up in nearly every article on course length and neither survives a check.

The eight second attention span

The claim that human attention span fell to 8 seconds, shorter than a goldfish, did not come from Microsoft's research. It appears in a 2015 Microsoft Canada consumer report, attributed to a website called Statistic Brain. When the BBC investigated in March 2017, none of the sources Statistic Brain listed could produce any research behind the figure.

Gemma Briggs, a cognitive psychologist at the Open University quoted in that piece, put it plainly: the idea of an average attention span is pretty meaningless, because attention is task dependent. The goldfish comparison is also unevidenced, and goldfish are used as a model organism precisely because they learn and retain.

If you want a real number about modern attention, Gloria Mark's research found average time on a single screen fell from about two and a half minutes in 2004 to about 47 seconds. That is about task switching in knowledge work, not a biological limit, and it is not an argument for 47 second lessons.

The completion rate by length table

You will see a table claiming courses under 2 hours complete at 70 to 85 percent, 2 to 6 hours at 60 to 70 percent, and 6 hours plus at 40 to 60 percent. It is attributed, where it is attributed at all, to research from major online learning platforms. There is no study, no platform named, no year and no link. It is recycled across dozens of articles.

The one defensible finding on length and completion is Jordan's 2015 regression across 121 MOOCs, which found each additional week cost about 1.75 percentage points, significant at p below .001. That is a real number about real courses, and it is much less dramatic than the fake table.

What length do courses actually run?

Class Central's analysis of Udemy's catalogue found the average course ran 3.79 hours in 2025 with a median of 2.25 hours, down from a 4.69 hour average in 2024 and 7.7 hours in early 2023. The median barely moved, so the average fell because of a flood of short courses rather than existing courses getting shorter.

The same analysis contains the finding that makes this a genuine trade off rather than a simple rule:

• Courses under 1 hour: about 2,100 average enrolments.
• Courses of 15 to 20 hours: about 7,300.
• Courses over 20 hours: about 11,400.

Ratings correlate positively with length too. Buyers reward length at the moment of purchase and learners punish it at the moment of consumption. On a marketplace where the sale is the whole game, long wins. For a coach whose reputation depends on clients getting a result, it does not.

Length is a production cost decision

The part missing from every article on this subject: runtime is what your course costs to build.

Editing is the dominant line item in course production, and it scales with finished minutes. Doubling your runtime roughly doubles the edit hours, doubles the re recording burden every time your offer changes, and doubles the delay before you can launch. A twelve hour course is a four times bigger production commitment than a three hour one, for a course whose median viewer will abandon it earlier.

Run the decision as arithmetic rather than instinct. Cost to produce a module, against the probability that module is ever watched. On the evidence above, module twelve of a twelve hour course has a low probability and a full cost. Cutting it is free money and better completion.

A workable default

• Total runtime: 2 to 4 finished hours for most coaching offers.
• Modules: 5 to 8, each mapping to one outcome the buyer can name.
• Lessons per module: 3 to 5.
• Lesson length: under 6 minutes where the material allows, up to 10 when it genuinely does not.
• Anything longer needs a specific reason, such as a certification or a technical curriculum where the length is the product.

Plan the segments before you film. Guo's clearest practical finding was that splitting long recordings afterwards does not recover the engagement of material planned short from the start.

Key takeaways

• Median engagement approaches full length below 6 minutes and falls to roughly 20 percent between 12 and 40 minutes.
• Maximum median engagement for a video of any length is about 6 minutes, so extra runtime is often unwatched.
• Every extra week of course length costs about 1.75 percentage points of completion.
• On Udemy, courses over 20 hours average 11,400 enrolments against 2,100 for courses under an hour. Long sells, short finishes.
• The 8 second attention span claim traces to a consumer report citing a site that could produce no research. Ignore it.
• The completion by length percentage table circulating everywhere has no source at all.
• Runtime is a production cost. Doubling finished minutes roughly doubles the edit hours and the launch delay.

Common questions

How many modules should an online course have?

Five to eight for most coaching offers, each mapping to a single outcome the buyer could name before purchasing. Module count matters less than whether each one earns its place. Every extra week of material costs roughly 1.75 percentage points of completion, so unnecessary modules are actively expensive.

How long should each lesson be?

Under six minutes where the material allows. Median engagement is close to the full length below six minutes and drops to roughly 20 percent for videos of 12 to 40 minutes. Plan lessons that short from the start, because splitting a long recording afterwards does not recover the effect.

Will a short course feel like poor value?

At the point of sale, possibly. Udemy data shows longer courses earn more enrolments and better ratings. The fix is not padding but specificity: name the outcome, show the path, and price on the result rather than the runtime. Padding costs you completion and production budget simultaneously.

Is the six minute rule still valid in 2026?

It remains the best available evidence, but it is 2014 data from free STEM MOOCs and has never been replicated on paid creator courses. Wistia's 2026 data suggests educational and long form content holds attention better than other formats. Treat six minutes as a strong default, not a law.

Deciding your runtime

Write the outcome for each module first, then delete any module that does not change what the buyer can do. Whatever survives, multiply the finished minutes by what editing costs, because that number decides whether the course is worth building at the price you were planning to charge.

EchoPulse produces course video for coaches and course creators, at a 48 hour standard turnaround per deliverable and with published pricing. The $299 Pilot runs 14 days, you keep everything produced, and there is no contract.