In this seminar, Tha o Le will give a talk about Growth in Saturated Market: A Markov-base d Network Optimization Approach.
In mature markets where t otal demand is fixed, competitive growth depends not only on attracti ng customers but also on the timing of their purchases. Yet most quantitative marketing models--whether probabilistic choice, state-d ependent demand, or network-based diffusion--capture which&n bsp;option consumers switch to but not when switching o ccurs, despite substantial heterogeneity in purchase timing driven by frictions, habits, and depletion dynamics.
We addr ess this gap by introducing a heterogeneous-time Markov network that jointly models probabilistic and temporal transitions in purchase beh avior. Methodologically, we derive closed-form expressions for market share, sales rates, and inter-purchase times in this heterogeneous-t ime environment, enabling analytical tractability without distributio nal assumptions or simulation. Building on these results, we develop a gradient-based optimization framework that respects fixed market de mand and identifies minimal, targeted interventions for shifting shar e. Using IRI panel data, we demonstrate that jointly optimizing switc hing probabilities and switching times yields significantly larger ga ins and requires smaller deviations from observed behavior than proba bility-only approaches.
Our work provides a m odern quantitative framework for understanding and optimizing competi tive dynamics in mature, zero-sum markets, and highlights the role of temporal heterogeneity as an underexplored driver of firm performanc e.
DESCRIPTION: In this seminar, Thao Le will give a talk about Growth in Saturated Market: A Markov-based Network Optimization Approach. In m ature markets where total demand is fixed, competitive growth depends not only on attracting customers but also on the timing of their pur chases. Yet most quantitative marketing models--whether probabil istic choice, state-dependent demand, or network-based diffusion--cap ture which option consumers switch to but not when< /em> switching occurs, despite substantial heterogeneity in purc hase timing driven by frictions, habits, and depletion dynamics. � � We address this gap by introducing a heterogeneous-time Markov network that jointly models probabilistic and temporal transitions i n purchase behavior. Methodologically, we derive closed-form expressi ons for market share, sales rates, and inter-purchase times in this h eterogeneous-time environment, enabling analytical tractability witho ut distributional assumptions or simulation. Building on these result s, we develop a gradient-based optimization framework that respects f ixed market demand and identifies minimal, targeted interventions for shifting share. Using IRI panel data, we demonstrate that jointly op timizing switching probabilities and switching times yields significa ntly larger gains and requires smaller deviations from observed behav ior than probability-only approaches. Our work provi des a modern quantitative framework for understanding and optimizing competitive dynamics in mature, zero-sum markets, and highlights the role of temporal heterogeneity as an underexplored driver of firm per formance. END:VEVENT END:VCALENDAR