Exploring the design and utility of an integrated web-based chatbot for young adults to support healthy eating: a qualitative study

Background There is a lack of understanding of the potential utility of a chatbot integrated into a website to support healthy eating among young adults. Therefore, the aim was to interview key informants regarding potential utility and design of a chatbot to: (1) increase young adults’ return rates and engagement with a purpose-built healthy eating website and, (2) improve young adults’ diet quality. Methods Eighteen qualitative, semi-structured interviews were conducted across three stakeholder groups: (i) experts in dietary behaviour change in young adults (n = 6), (ii) young adult users of a healthy eating website (n = 7), and (iii) experts in chatbot design (n = 5). Interview questions were guided by a behaviour change framework and a template analysis was conducted using NVivo. Results Interviewees identified three potential roles of a chatbot for supporting healthy eating in young adults; R1: improving healthy eating knowledge and facilitating discovery, R2: reducing time barriers related to healthy eating, R3: providing support and social engagement. To support R1, the following features were suggested: F1: chatbot generated recommendations and F2: triage to website information or externally (e.g., another website) to address current user needs. For R2, suggested features included F3: nudge or behavioural prompts at critical moments and F4: assist users to navigate healthy eating websites. Finally, to support R3 interviewees recommended the following features: F5: enhance interactivity, F6: offer useful anonymous support, F7: facilitate user connection with content in meaningful ways and F8: outreach adjuncts to website (e.g., emails). Additional ‘general’ chatbot features included authenticity, personalisation and effective and strategic development, while the preferred chatbot style and language included tailoring (e.g., age and gender), with a positive and professional tone. Finally, the preferred chatbot message subjects included training (e.g., would you like to see a video to make this recipe? ), enablement (e.g., healthy eating doesn’t need to be expensive, we’ve created a budget meal plan, want to see?) and education or informative approaches (e.g., “ Did you know bananas are high in potassium which can aid in reducing blood pressure?” ). Conclusion Findings can guide chatbot designers and nutrition behaviour change researchers on potential chatbot roles, features, style and language and messaging in order to support healthy eating knowledge and behaviours in young adults.