Let’s have a chat We’re all talking about it! The launch of open AI platforms like ChatGPT has caused much excitement, but latterly, also some significant concern. But in the midst of this wider and important debate about safety and legislation, let’s not forget the value of more bespoke and tailored AI solutions which already support us as market research professionals in our day-today lives. Specifically, HRW recently conducted a self-funded study, ostensibly designed to develop a new quantitative approach to concept testing, but also incorporating a chatbot tool; a conversational AI companion who prompts respondents for more information based on the answers to the questions we pose within an online survey. There are various ways in which a chatbot can be incorporated into market research surveys, but for this project we chose to simply use the chatbot in place of standard open-ended questions, allowing us to engage respondents in a mini conversation to provide fuller and more robust answers to our questions. The chatbot output certainly provides an extra level of understanding. And here’s why: In a standard open-end, we gain top level responses. Important initial reactions but with little supporting insight. The chatbot gives far more depth and detail Specifically, the AI is programmed to detect key words and phrases identified as being important and will pull out these responses to adapt and refine the conversation Additional analysis approaches are vast. With a complete and visual analysis dashboard, it’s possible to drill down to draw out key themes, conduct sentiment analysis, identify differentiating terms, frequently mentioned themes and specific language (all without the laborious process of coding) Fig 1: Example conversations Fig 2: Example sentiment analysis As with all approaches, this additional insight comes with its own challenges. For example, we initially encountered some programming issues relating to survey redirects, but this was resolved and is a solution we can now readily employ. It’s also important to consider the time to set up the AI model, which requires human input to develop focus terms with which the chatbot will be programmed. This particular survey focused on anti-smoking campaigns and so we were intuitively able to consider how respondents might react and the features of the adverts tested that we knew to be important. In more complex scenarios and considering the highly technical language used by HCPs, input from medics may well be beneficial to fine-tune the programme to ensure we capture the most meaningful results. But despite these additional considerations, this particular form of AI is certainly an approach worth considering. More and more in market research are we seeking hybrid methodologies which allow us to maintain the robustness afforded by quantitative approaches but melded with our understanding of the whys. The chatbot allows us to do just this, providing a great mix of quantitative and qualitative insight and importantly, in an engaging format for survey participants, potentially fatigued by dry and often lengthy research surveys. Watch this space for more information about the self-funded study referenced in this article to see how a new quantitative approach can support the decisions you make when developing concepts By Nicola Vyas and Imogen Chandler Apply Now!