Estimating Optimal Recommendation Policy Under Heterogeneous Treatment Effect of Product Recommendation
The effect of collaborative filter-based product recommendations ‒ a popular tool for optimizing online sales – can widely vary across products with different attributes. While the product features and their interactions can have a highly nonlinear influence on the effect of recommendations, academic studies have only examined their linear variations on individual product features. We utilize a nonparametric causal forest model to estimate the nonlinear variations in the effect of product recommendations with combinations of different product feature values on the field data of a mid-size fashion e-tailer in the US. Under certain combinations of product feature values, we find a negative and significant effect of recommendations suggesting that it is optimal to not offer recommendations in such cases. We estimate a policy learning tree that provides optimal recommendation policies under different combinations of product feature values. We further show that the e-tailer can obtain up to 5.4 percent higher sales by following the optimal recommendation policies. Our research has implications for optimizing product sales with algorithmic product recommendations
Year of publication: |
[2022]
|
---|---|
Authors: | Wan, Xiang (Shawn) ; Kumar, Anuj ; Aytug, Haldun |
Publisher: |
[S.l.] : SSRN |
Saved in:
freely available
Extent: | 1 Online-Ressource (37 p) |
---|---|
Type of publication: | Book / Working Paper |
Language: | English |
Notes: | Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments February 16, 2022 erstellt |
Other identifiers: | 10.2139/ssrn.4036840 [DOI] |
Source: | ECONIS - Online Catalogue of the ZBW |
Persistent link: https://www.econbiz.de/10013294852
Saved in favorites
Similar items by person
-
Kumar, Anuj, (2020)
-
Kumar, Anuj, (2019)
-
What Doesn't Kill You Makes You Stronger? Evidence from Vampire Attack on Decentralized Exchange
Zhao, Xi, (2023)
- More ...