Ask ten course creators how long a lesson should be and you'll get ten different opinions, most of them anchored to whatever length the creator personally finds comfortable to record rather than what the completion data on video platforms actually shows. The honest answer is that length itself isn't the enemy, a video can run 45 minutes and still get watched to the end if every one of those minutes is doing something, but past a certain point most lessons stop pulling their weight, and it helps to know roughly where that point tends to sit before you build your course around the wrong instinct.
What the drop-off curve on long videos actually looks like
Video engagement doesn't decline in a straight line, it holds fairly steady for the first several minutes and then starts falling away faster the longer a video runs, a pattern that shows up consistently across video hosting platforms regardless of subject matter. A student who commits to pressing play has already decided the topic is worth their time, so the first few minutes rarely lose anyone. What changes their mind partway through is usually one of two things: either the lesson has drifted off the specific task it promised and into tangents, or the student's attention has simply run its natural course for a single sitting and the video keeps going past it. Once a lesson crosses roughly the 15 to 18 minute mark, you'll typically see completion fall off noticeably compared to shorter lessons covering an equivalent amount of material, which is worth knowing before you record a 40-minute module and wonder later why your completion rate sits lower on that one lesson than everywhere else in the course.
The case for 6 to 12 minute lessons
The reason 6 to 12 minutes keeps coming up as a sweet spot isn't magic, it's that most single, well-scoped skills genuinely fit inside that window once you cut the throat-clearing at the start and the recap at the end. A lesson that opens with "in this video we're going to talk about" and closes with "so to summarize everything we just covered" is padding a real six-minute lesson out to nine minutes of mostly filler, and cutting that filler doesn't just save time, it raises the ratio of useful content to total runtime, which is what actually keeps someone watching. Shorter lessons also make the rest of your course structure work better, because a student choosing between finishing one short lesson before bed and starting a 35-minute module will pick the shorter option almost every time, and choosing to start is most of the battle in getting a course finished at all.
When a longer deep-dive earns its length
None of this means every lesson has to be short. A genuine deep-dive, walking through a full client project start to finish, or working through a complex case study where the value is in seeing the whole process connected rather than chopped into pieces, can justify 25 or 30 minutes if the student arrived at that lesson already invested and the content stays tightly on-topic throughout. The test isn't the clock, it's whether you could split the video into two or three separate lessons without losing anything, and if the answer is yes, you should probably split it regardless of how the length looks on paper. If the answer is genuinely no, because the whole point is watching one continuous process unfold, then a longer lesson placed later in the course, after a student has already built momentum on shorter wins, tends to hold attention far better than the same length placed as lesson one.
There's also a difference between a deep-dive that stays long because it needs to and one that stays long because nobody trimmed it. A useful gut check is to record the lesson, sit down the next day, and watch it back at 1.5x speed the way an impatient student might. If you catch yourself reaching for the skip-forward button even once, that's the section a real student will skip too, and it's worth cutting before you publish rather than after your completion numbers tell you the same thing three weeks later.
How to test your own video length, not someone else's rule
The ranges above are a starting point, not a law, and the only way to know what's right for your specific course is to look at your own numbers. If you're using course hosting that shows lesson-by-lesson completion, find the video where the drop-off is sharpest relative to the lessons around it and ask honestly whether it's dragging, wandering off the stated topic, or just genuinely long for what it covers. Try splitting that one lesson into two and watch what happens to completion on both halves over the next batch of students. Small, single-lesson experiments like that tell you far more about your specific audience and topic than any general benchmark, including the ranges in this article, ever will.
Give any change like this a real sample size before deciding it worked, since a handful of new students isn't enough signal to tell a genuine improvement from ordinary week-to-week noise. Wait until at least twenty or thirty students have gone through the reworked lesson, then compare its completion rate against the same lesson's old numbers rather than against a different lesson entirely, because two lessons on different topics were never going to have identical completion rates to begin with.
Length is a tool for keeping a promise, the promise your lesson title made when a student clicked play. Whether that promise takes four minutes or twenty-four to keep, the video should end the moment it's kept and not a beat later, and building that discipline into how you record is worth more to your completion numbers than any specific minute count you memorize from a blog post.