This article discusses the comparative effectiveness of student input at different times during the design of CALL tools for learning kanji, the Japanese characters of Chinese origin. The CALL software "package" consisted of tools to facilitate the writing, reading and practising of kanji characters in context. A pre-design questionnaire was analysed to determine what students perceived would be effective software tools for kanji learning. This was followed by a written process tracing protocol to ascertain if the CALL tools, created by the author and the technology designer were effective in meeting students' needs. Based on a detailed analysis of students' responses in the first two stages, changes were made to the CALL tools interface and the learning environment. The final stage of student input was a post-design evaluation completed by a further cohort of students, focusing on whether the data collected previously had produced changes that were transferable and could effectively meet the learning needs of different groups of students. All three stages of input produced different data of importance to the final design. Mid-design input was the most detailed and transferable design elements could be extracted. My recommendation is that end-users of CALL tools be consulted throughout the design process.