This is the question people ask right before they record, and it's usually standing in for a different one: how do I know if this lesson is any good? Length is a proxy, but it's a useful one, because a lesson that runs long is nearly always a lesson doing two jobs.

One lesson, one job

Write down what the student can do at the end of the lesson that they couldn't do at the start. If you need the word and to write it down, you have two lessons.

That single test does more for course quality than any rule about minutes. A lesson with one job has an obvious beginning, an obvious end, and an obvious action at the end of it — which is also what makes it feel finishable.

Under ten minutes is a constraint, not a law

Aim under ten. Not because attention spans are collapsing, but because a ten-minute ceiling forces the edit: it removes the throat-clearing intro, the tangent about the time a client did something similar, and the recap of the previous lesson that nobody needed. What survives is the teaching.

There are real exceptions. A live-build walkthrough, a full teardown, a worked example where the value is watching the whole thing happen — those can run long and should. What can't run long is explanation.

Signs your lesson is actually two lessons

  • You say "and the other thing to know here is" and change subject.
  • The action at the end has two parts that don't depend on each other.
  • You recorded it twice because you kept losing your place in the middle.
  • You can't name it without a colon or a slash.

What to do with the twenty-minute lesson you already have

Don't re-record it. Split it:

  1. Watch it and mark the point where the subject changes. There is almost always exactly one.
  2. Cut there. You now have two lessons, and the second one starts on a natural beat.
  3. Give each half its own title and its own action.
  4. If the halves are unequal, that's fine — a four-minute lesson is not a failure.