Automated feedback in GeoGebra: collaborative teacher design and primary pupils' responses

[Objective] The study aimed to document how a researcher and a primary-school teacher, working within an online teacher-education project, designed a GeoGebra task with automated feedback, and to examine how pupils aged eight responded to it. [Methodology] A qualitative case study with action-research features was conducted in two phases. From September 2023 to June 2024, teachers and researchers from Brazil, Portugal and Cape Verde met monthly online to design tasks with automated feedback. On 27 February, 12 March and 13 March 2025, three pairs of pupils from a third-year class in a public school in Porto used the resulting bakery applet, in which exactly 30 coins had to be spent on loaves and baguettes. Session transcriptions were analysed with the design records and the applet file. [Results] The applet combined a message reporting the total cost of each entry, coloured rows of coins, a counter of failed attempts and input restrictions. Depending on the randomly drawn prices, the task admitted between three and sixteen exact solutions. One pair revised its strategy using the feedback and solved three challenges. One pair solved the first challenge, then hesitated and did not converge. One pair erased entry without reading the messages and did not solve the challenge. [Conclusions] Feedback reporting the consequence of an answer supported revision only when pupils could read the statement, calculate with the prices and accept the risk of error. Design should anticipate these conditions. The findings concern six pupils and one task.