EXPRESS: Amplifying Insights: Harnessing Voice Technology to Decode Customer Uncertainty in Marketing

Verbal responses are widely collected in marketing through voice surveys and phone calls. Recent advances in voice technology enable firms to go beyond textual content and extract information from voice, such as speaker identity and sentiment. However, little is known about whether voice can be used to infer response uncertainty, a key construct for understanding and reducing bias in customer feedback. Existing marketing measures of uncertainty were developed largely for selected-response surveys and face limitations in cognitive simplicity, predictive robustness, and applicability. We examine two methods that infer uncertainty as expressed in vocal delivery: one based on raw audio data, and the other based on a voice feature (i.e., pitch). We evaluate both methods in a laboratory net promoter score (NPS) study and a field NPS study conducted with a full-service music store chain; the voice feature approach outperforms the raw audio approach in marketing applications while preserving privacy. The collaborating firm suggested two managerial benefits of inferring response uncertainty: improving incentive program effectiveness by targeting customers hypothesized to be more likely to redeem coupons, and identifying brand ambassadors who provide persuasive testimonials. These firm-driven insights inform an actionable framework for using voice-inferred uncertainty in marketing practice.