The Interplay Between Customer Feedback Solicitation and Innovation: A Dynamic Solution

Agile product development relies on rapid iteration and customer feedback to guide product improvement, yet firms face challenges in coordinating when to solicit feedback and when to invest in quality improvement. While each action adds value on its own, coordinating the two introduces an additional layer of complexity beyond optimizing either decision in isolation. We model this problem as a semi-Markov decision process, with product quality and accumulated customer feedback as state variables. To build intuition, we first analyze two single-action settings: one where the firm continuously invests in quality improvement while optimizing feedback solicitation, and another where it always solicits feedback while optimizing quality investment. In both cases, the optimal policy exhibits a clean monotone-threshold structure. We then study the full joint optimization problem, where the firm must simultaneously decide when to solicit feedback and when to invest in quality. These two interdependent actions create significant analytical complexity. Despite this, we uncover a key structural insight: under mild regularity conditions, the optimal joint policy preserves the threshold-based properties of the single-action settings. Feedback solicitation follows an optimal stopping rule, while quality investment is selectively applied based on the evolving product state. This structure partitions the state space into four regions, each guiding a distinct course of action. We also conduct sensitivity analysis to show how the thresholds shift in response to changes in model parameters. Our analysis reveals that effective agile product development benefits from selective agility—instead of defaulting to nonstop iteration, firms may benefit from pausing quality investment to await more feedback, or focusing solely on quality improvement once sufficient customer input has been collected. The resulting threshold-based policy offers clear, actionable guidance for balancing learning and execution under uncertainty.