Chapter-level metrics give you a slightly finer grained measure of where problems exist. Use the “sort” feature to sort by each metric. Pay attention to these things:
Avg Rating
This is the primary metric for learner satisfaction with the chapter. Although it doesn't tell you why learners dislike the content, it can be used to determine where you should focus your maintenance attention.
A dip in a Chapter’s Average rating can indicate structural problems with that chapter. For example, is the topic dull or not solving real-world problems? Is the dataset boring? Does the flow of exercises not make sense? You'll need to look closer at the feedback messages in the exercise-level metrics to determine what went wrong.
% Completions
This is the primary metric for learner engagement.
Typically completion rate is lower for Chapter 1 (because students start the course out of curiosity, or after seeing an advert, then drop out), then stabilizes from Chapter 2. If you see a big drop in this metric in the final chapters, there is an engagement problem. Are there exercises that are impossible to complete? Is the content boring?
Number of Messages
This is the primary metric for mistakes in content, and unclear or annoying content.
The count refers to the messages generated via the "Report Issues" tool in the exercise interface. See the exercise-level metrics to determine the exact location of the problems.
% Hint and % Solution
These are the primary metrics for the difficulty-level of content.
These metrics are calculated as the proportion of exercise starts where a hint or solution was used. Ideally, a course should have a gradual increase in the level of difficulty as it progresses. That is, these numbers should gradually increase from Chapter 1 to the final Chapter. Since difficulty-level can vary a lot from exercise to exercise, it is vital to inspect the % asked hint and % asked solution metrics on the exercise-level to determine what needs maintaining.
A % Hint above 20% is an indicator that there may be an issue, above 35% is a red alert that something needs to be addressed. For % Solution, a rate above 20% is a sign that something needs to be fixed.