Most people who buy a sixty-hour course never finish it. That isn't a discipline problem — it's a design problem. A course is a commitment device, and the bigger the commitment, the more likely it collapses the first week life gets busy.
The completion cliff
Look at any large cohort and the drop-off is brutally consistent:
- Module 1: nearly everyone shows up
- Module 3: about half remain
- Module 8: a stubborn, self-selected fifth
The learners who leave aren't less capable. They simply lost the thread — a week away from a monolithic course means re-reading three hours of context before you can write a single line of code again.
Small units, real endings
A micro-course does one thing: it ends. Six to twelve hours, one skill, one artifact you can point at. That ending matters more than the content, because a finished thing produces momentum and an unfinished thing produces guilt.
Momentum compounds. Guilt compounds faster.
Stacking beats sprawling
Once units are small, they become composable. Pandas is a module. So is NumPy, so is Matplotlib. Stack those three with Excel and Tableau and you have an analyst track — assembled from parts that each stood on their own.
This gives you two things a mega course can't:
- Reordering. Already know SQL and Excel? Skip them. Your path shortens instead of forcing you through a fixed syllabus.
- Reuse. Python appears in the AI track, the ML track and the automation track. You learn it once and it counts everywhere.
How to sequence your own stack
Pick the role first, then work backwards:
Role -> Track -> Micro-courses -> This week's module
Never start from the module. Starting from the module is how people end up with eleven half-finished tutorials and no job change.
The practical rule
Choose the smallest unit that still produces something you'd show another engineer. If a module can't produce that, it's too small. If you can't finish it inside two weeks of ordinary life, it's too big.