When to Stop a Course and Start Building
Courses are a means to an end. Once you can explain the core ideas and need practice, more lectures often slow you down.
On this page
The point at which a course stops helping is rarely announced. Lectures keep arriving, the next module looks sensible, and the sense of progress comes from watching content rather than producing anything. Knowing when to stop, and what to build instead, is a skill of its own.
Signs that you have enough for now
There are reliable signs that a course has given what it can. You can state the core ideas of the current module without notes. You can explain why a method works and name a case where it fails. When a lecture starts, most of its content is familiar, and what remains new is detail that could be looked up when needed.
A direct test is more informative than a feeling. Close the course, pick a small task that uses the material, and attempt it. The points where you stall show the real gaps. If the gaps are narrow and can be filled by looking up a reference, the course has done its job. If they are wide and basic, the course is still useful, and the better next step is more practice within it. Boredom is a weaker sign than it seems. A course can feel easy for the parts already known and hard for the parts not yet learned, so the test is the task, not the mood.
Why lectures start to slow you down
Courses usually front-load ideas before application. That order suits a beginner, but it has a cost. After a point, each new lecture adds less than a single problem that forces a decision. A problem asks you to choose a method, justify it and check the result. A lecture asks you to follow an argument someone else has already made, and the argument can be complete and clear while your own grasp of it stays thin.
The shift is gradual. Watch for lectures you could summarize before they finish, or exercises you can complete by pattern without knowing why they work. Those are signs that the next lecture will teach less than a project would. A lecture on a method may take most of an hour to show what a single afternoon of implementing it would make clear, and the afternoon is where the gaps appear. This does not make lectures useless. They are efficient for definitions, and for the reasons behind a result that would take a long time to reconstruct alone.
Choose the next project carefully
A good next project uses the most recent idea, works with a dataset or problem you can obtain without difficulty, and is small enough to finish in about a week. Introduce one new concept per project. If the last module covered classification, build a classifier for a folder of your own notes labeled by topic, and keep the rest of the setup simple. The aim is to meet the idea again in a situation the course did not design. A project that sits too close to the course can feel like homework, and one that sits too far away can stall. Aim for the middle, where the new idea is clear but the path to a result is not yet obvious.
Avoid projects that depend on many new tools at once. When something fails, you want to know which part caused it. Keep the moving parts to one new thing, and let the familiar parts stay familiar. If the project grows past its week, cut the scope rather than extending the time. The finish line is what makes the next project possible.
Use the course as a reference
Once building has started, keep the course open as a reference rather than a sequence. When a specific question arises, go to the section that answers it. This changes the relationship with the material. A lecture is watched once, in order, while a reference is consulted when a question needs answering. The second use often teaches more, because the question is one you actually have.
It also makes returning to the course easier. You will know which sections you needed and which ones you can skip.
Guard against collecting completions
Completion certificates and finished modules feel like progress. They are easy to count and easy to share. The risk is that someone can finish many courses and still be unable to build something small alone. A useful check compares two numbers: the courses finished and the things built and explained. If the first number grows much faster than the second, the pattern is worth changing.
The check is not a reason to avoid courses. It is a reason to stop treating each completion as the goal. Some courses are still worth finishing, especially when they give access to people, data or tools that a project needs. The question is whether a course is serving a project or standing in for one. Count only the things another person could inspect, such as a repository, a written explanation or a working tool, since those are the outputs that show skill.
Deciding what to build next
List the three most recent lessons completed. For each, write one question you could not answer without looking it up. Choose the project that requires answering those questions, and define it using the rule from earlier: one new concept, a dataset you can get, and a finish line about a week away.
Set a limit on the course for the project's duration. The course may still be consulted as a reference, but no new module should start until the project is finished or deliberately set aside. At the end of the week, write down what the project required that the course did not cover. That list is the most reliable guide to what to study next, and it comes from your own work rather than from a syllabus.
