A score can show that a student missed a question. It cannot, by itself, explain why. A session log can show time spent on a task, but not whether that time was productive, confusing, or interrupted.
Learning data is most useful when it leads to a better question.
Start with a teaching decision
Decide what you need to know before opening a report. Are students ready for a new topic? Is the same misconception appearing in several responses? Does one learner need a different explanation?
Collect only the information relevant to that decision. More numbers do not automatically create a clearer picture.
Look at the work behind the signal
If several students miss a fraction question, compare their working. One may be adding denominators; another may have misunderstood the wording. Those students may need different follow-up questions even though their scores match.
Ask the learner what they were thinking. Treat an activity pattern as a clue, not a diagnosis, a fixed ability label, or proof of motivation.
Make one change and check again
Try a different worked example, clarify an instruction, or give a focused practice question. Then look at the next attempt. Keep your judgment tied to what the student can explain and do, not only to a dashboard indicator.
Keep visibility proportionate
Use school-approved systems and follow your school's rules for access, retention, and sharing. Do not place identifiable student work in public reports or general-purpose AI tools.
For families using ScholarTree, learning insights sit alongside full chat replay and voice replay with audio. Those conversations can add context, but they do not replace a teacher's assessment or a parent's conversation with the child.
The goal is not to measure everything. It is to choose a more helpful next step.