Your knowledge base is already most of the course
The expensive part of training is usually not writing it. It is writing it again, in a different shape.
Part of a single argument: the engagement gap is a workload problem.
Most organisations that struggle to produce training are not short of material. They are drowning in it. Policies, decks, recorded calls, help centre articles, the long Slack message someone wrote in March that explains the thing better than any official document.
What they are short of is a route from that material to something a learner can follow. So the material sits where it is, and somebody starts a fresh document, and writes down what the company already knows.
The gap is shape, not substance
A policy document and a lesson contain much the same facts. They differ in arrangement, in what they assume, and in what they are optimised for.
A policy is written to be complete and defensible. Every case is covered, the ordering follows the logic of the rules, and nothing is emphasised because emphasis implies the rest matters less. It is a reference document, and it is a good one.
A lesson is written to be understood in sequence by someone who does not yet know the material. It has an order, an emphasis, and a point at which it stops. Those are editorial decisions, and they are the actual work.
This distinction matters because it tells you what is genuinely being produced when someone turns the first into the second. Not new knowledge. A new arrangement of existing knowledge, which is a much smaller job than the blank page makes it feel.
Why it still gets rewritten from scratch
Three reasons, all mechanical.
The source is hard to find, so starting fresh is more predictable than searching. The source is hard to trust, because nobody is sure which version is current. And the transformation is tedious enough that once someone has opened a blank document, continuing is easier than going back to look.
None of these is about the difficulty of the writing. They are all about the state of the material around it, which is a different problem and a considerably more tractable one.
What changes if the source is the course
If lessons are built from the knowledge base rather than beside it, three things follow.
Updating the source updates the training, or at least flags the lessons that need attention. The two stop drifting apart, which is the failure mode where the onboarding module and the actual policy quietly disagree for eighteen months.
The expensive editorial judgement gets spent once and reused, rather than being re-made by each person who needs to teach the topic.
And the cost of a new course falls as the knowledge base grows, which inverts the usual relationship where the tenth course costs the same as the first.
What still needs a person
Deciding what matters. A source document treats everything as equally important because it must; a course cannot, and choosing is judgement.
Knowing what your particular learners get wrong. This is not in any document. It lives with whoever has taught it before and watched people misunderstand the same thing repeatedly.
Writing the check that discriminates. Turning a fact into a question that separates the people who understood it from the people who recognise the words is genuinely difficult and genuinely valuable.
That is a real afternoon of expert time and it should be protected. Everything around it, the finding, the reformatting, the retyping, was never expertise and should not be charged as if it were.
Where to start
Take the course you have rebuilt most often. Find the source it was built from. If you cannot find the source, that is the finding, and it is more important than anything else in this post.
If you can find it, look at how much of the course is arrangement of that source and how much is genuinely new. In most cases the ratio is uncomfortable, and it is the clearest available argument for changing how the two are connected.