In recent years, the problem of the number of students enrolled in STEM courses falling below the minimum required to carry out the courses has worsened. This situation, which used to occur primarily in degree programs at the end of their life cycle, is increasingly happening in active STEM degree programs as well. The declining attractiveness of traditional STEM degree programs and negative demographic trends are seen as the main causes of this problem. To assist students and their teachers in this situation, a novel approach to grading student work based on the level at which students achieve learning outcomes in an online system was explored. Since the focus was on STEM students, math-based tasks were used to assess student knowledge within the online system. In addition to the accuracy of the final results of the tasks, the accuracy of the intermediate results was also monitored. Each result was linked to the corresponding concepts, which in turn were linked to the corresponding learning outcomes. By entering correct intermediate and final results, the online system was able to monitor student success in mastering concepts and learning outcomes. The implemented online system utilizes fuzzy inference to calculate the levels at which students have achieved the learning outcomes. These levels are continuously recalculated during the semester and presented to the students through generated recommendations after each solved task. The proposed approach has positively impacted students’ motivation to learn during the semester, as confirmed by an anonymous questionnaire. Additionally, the results have shown that the final level at which students have achieved the learning outcomes within the online system, calculated at the end of the semester, can be used for grading and is comparable to the grades students would achieve in traditional midterm and final exams, thus helping teachers with their workload.