All interviews were audio-recorded and transcribed verbatim by a research assistant and a professional audio transcription company, and all personal identifying information was removed from the transcripts. Transcripts were imported in Atlas.ti (ATLAS.ti Scientific Development GmbH, Berlin, Germany). The process of coding was informed by the Qualitative Analysis Guide of Leuven (QUAGOL) [23], and was based on content analysis using a combination of inductive and deductive coding. First, the transcripts were read and re-read in order for the first author to become familiar with the data. This phase included underlining relevant passages and noting emerging themes in the margin of the transcript. In the second phase, the key elements of the transcripts were summarized in ‘narrative interview reports’, which aimed to holistically capture the experience of the participant. In the third phase, key elements that emerged from the transcripts and narrative interview reports appearing to be relevant for answering the research questions were translated into more abstract concepts. These emerging concepts were used to construct the initial codebook (e.g., inductive coding). For the first research question, this inductive approach was complemented with a deductive approach, by also including relevant concepts identified from literature (e.g., deductive coding) [24,25]. For the second research question, the codebook only included concepts that emerged from the narrative interview reports, thus following a primarily inductive coding approach, as research and theory concerning perceptions, needs, and preferences on nudging interventions are scarce. After coding the first five interviews, the codes and quotations were carefully examined and compared by the first and second author, and minor revisions to the codebook were made. The following five interviews were coded using the updated codebook, and the first five interviews were again analysed in order to adjust the coding following the revisions to the codebook. This iterative process of comparing codes and quotations among the first and second author was repeated until all interviews were analysed. Codes were structured into intrapersonal, interpersonal, socioeconomic and environmental determinants of food choice, following socio-ecological models of health [24]. In order to provide an indication of the relative importance of the identified determinants of food choice, the narrative report of findings is accompanied by a bar graph, indicating the number of interviewees mentioning the specific determinant (interview frequency), and the frequency of coding this determinant in the interview reports (quotation frequency).