The use of online digital environments at Jorge Tadeo Lozano University has allowed the implementation of a question bank, with feedback, with the purpose of supporting students in their learning process for the subject of Mathematical Thinking; this subject is taught through lectures and workshops that are interrelated and leaded by different professors, one lecture session and two workshop sessions per week. A pedagogical innovation project was developed along two academic periods with the purpose of observing if the use of this online question bank with feedback improved the appropriation of mathematical concepts. The questions that are part of this bank were designed following recommendations for elaborating multiple choice questions with one answer; the selection of distractors for each question was made following conceptual or procedural reasons, which is where most of the difficulties for students are when solving an evaluation on this subject. For each question a detailed description was made, including conceptual aspects and key features such as type of question, question statement, answer choices, reasons for each choice, key information that must be included in feedback (graphics, concepts, solved examples, etc.), cognitive and conceptual domain, level of difficulty, date of elaboration and revision, name of who elaborated the question and name of who reviewed it. Each question was made available on this bank, located at the Virtual Classrooms System at Jorge Tadeo Lozano University (AVATA is the name of this Learning Management System working over Moodle) and corresponding to the Mathematical Thinking course, organized on folders separated by topics and levels of difficulty, according to the available syllabus of the subject. Evaluation models that gathered the topics covered on each of the three segments that form the academic period were used; such formative assessment was made available for students two weeks before each course evaluation. Due to the features of this evaluation, online, with multiple attempts and accessible from multiple devices, students were able to prepare their evaluations with a higher level of autonomy and at the pace, they would establish. Surveys, interviews and focus groups from students and professors that were part of the Mathematical Thinking course were carried out, with the purpose of identifying qualitative features that could eventually lead to adjustments and improvements to the question bank, although not many students participated in these interviews and surveys. Among the qualitative aspects that were highlighted is the fact that students felt comfortable when using this kind of evaluation, as they did not feel the pressure of the results; they also pointed out the role of workshop professors as a motivating element to consult the question bank in order to prepare the evaluations and strengthen mathematical concepts. The data obtained from each evaluation model and the results from students during the academic periods were analyzed statistically, allowing to conclude that there is a significant difference on course grades, both partial and finals, for students who used the evaluation models and students who did not. Results, both qualitative and quantitative, indicated that in general terms there is an improvement in performance for students who access to these online evaluation models compared to students who don't. In order to continue progressing on the work presented here, it is necessary to feed the bank with more questions, considering the aspects indicated from students and professors and also as a way of expanding the possibilities of randomizing the questions as students use the bank. It is also important to carry out statistical analysis to questions, with the purpose of establishing with better precision their levels of difficulty and the levels of difficulty of the evaluation models themselves. [ABSTRACT FROM AUTHOR]