A system is configured to provide a survey interface that collects response data, including both quantitative and qualitative response data, using multiple capture mediums. Mediums used to capture response data include input forms that collect structured response data on particular questions, as well as multimedia input forms that capture and collect free form multimedia response data in video form. This mix of quantitative and qualitative response data is analyzed across multiple modalities and used to develop an indexed response dataset, which may be queried to determine a set of pre-configured insights. An insight interface visualizes these pre-configured insights and accepts additional queries to provide a query interface that draws from the static indexed response dataset to allow for dynamic, conversational querying for additional insights.