Digital Commerce and the Digitally Connected Enterprise
Loading...
Paper Type
Short
Paper Number
1682
Description
The low cost of digital experimentation and increasing capabilities of machine learning algorithms have opened up new avenues for personalization in online retail. In this project, we describe how firms can combine these techniques to find an optimal targeted discount strategy. We cast this task within a simple decision-theoretic framework, solve for the optimal solution, and describe how to use causal machine learning methods and data from an online experiment to estimate the requisite model parameters. To validate our methodology empirically, we apply it to data from a randomized experiment at an online retailer. We demonstrate that our proposed targeted discount strategy can be estimated using real-world data with sufficient accuracy to result in increased profits. By accounting for the discount rate and heterogeneity in both baseline response rates and treatment effects, our proposal significantly outperforms existing price-agnostic targeting practices.
Recommended Citation
Miller, Alex and Hosanagar, Kartik, "Personalized discount targeting with causal machine learning" (2020). ICIS 2020 Proceedings. 7.
https://aisel.aisnet.org/icis2020/digital_commerce/digital_commerce/7
Personalized discount targeting with causal machine learning
The low cost of digital experimentation and increasing capabilities of machine learning algorithms have opened up new avenues for personalization in online retail. In this project, we describe how firms can combine these techniques to find an optimal targeted discount strategy. We cast this task within a simple decision-theoretic framework, solve for the optimal solution, and describe how to use causal machine learning methods and data from an online experiment to estimate the requisite model parameters. To validate our methodology empirically, we apply it to data from a randomized experiment at an online retailer. We demonstrate that our proposed targeted discount strategy can be estimated using real-world data with sufficient accuracy to result in increased profits. By accounting for the discount rate and heterogeneity in both baseline response rates and treatment effects, our proposal significantly outperforms existing price-agnostic targeting practices.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.