One library, every audience: courses that get reused
The same knowledge has to reach four groups who need different things from it. That does not require four courses.
Part of a single argument: the engagement gap is a workload problem.
Most organisations teach the same material to several audiences. New starters, existing staff after a change, customers, partners. The standard response is a course for each, and it is the most expensive available option.
It is also the one that ages worst, because now the same fact lives in four places and someone has to remember all of them when it changes.
Vary the framing, not the facts
The mechanism of a thing is the same regardless of who is learning it. What differs is why they care, what they will do with it, and how much depth they need.
So write the mechanism once and let the audience-specific part be the framing around it: the introduction that says why this matters to you, the examples drawn from that group's work, the depth setting that includes or omits the advanced lessons.
This is the difference between four courses and one library with four paths, and it is roughly a quarter of the maintenance.
Make lessons stand alone
A lesson that depends on three earlier ones can only ever appear in one sequence. A lesson that names its prerequisite and is otherwise self-contained can be dropped into any path.
Independence is what makes reuse possible, and it costs a little at authoring time: you have to resist the natural tendency to write continuously, referring back to what you said a moment ago.
The payoff is that a new audience becomes a matter of selecting and ordering existing lessons rather than writing a course.
Where audience-specific content is genuinely needed
Not everywhere, but not nowhere either. It is worth being precise about the exceptions.
Regulatory content sometimes differs by jurisdiction or role in ways that cannot be handled by omission. Customer-facing material may need different terminology for the same concept. And the examples, which carry most of the learning, usually do need to be drawn from the learner's own work to land.
Examples are the interesting case: they are the part that most needs to vary and the part most often left generic, because varying them is work. It is work worth doing, and it is much cheaper when the surrounding lesson is shared.
The failure mode of shared libraries
Over-generalisation. In the effort to write once for everybody, teams produce material that is technically applicable to all audiences and useful to none, because every concrete detail has been sanded off.
The guard against this is to write the mechanism for the audience that needs it in most depth, and let the others take a subset. A shortened version of something specific reads better than a generalised version of everything.
What this adds up to
One place where each fact lives. One editorial judgement about what matters, made once by whoever knows best. Paths assembled per audience rather than courses written per audience.
The result is that the marginal cost of teaching a new group falls to a selection exercise, and the marginal cost of a change falls to editing one lesson. That is the point at which training stops being a project each time and starts being something the organisation simply does.
Which is where this blog has been heading throughout. The barrier was never willingness. It was the amount of work, and most of that work was the same work, done again.