Working Papers
- “Price Promotions and Online Product Evaluations” (2026). Yongdong Liu, Yiting Deng, Anja Lambrecht, and Zihao Zhou. [Link]
[Abstract]
Digital products, such as mobile apps, e-books, and online courses, often have near-zero marginal costs of production and distribution. This enables frequent and steep price promotions. While it has been shown widely that price promotions increase sales, at least in the short term, the effect of price promotions on consumers’ product evaluations is less clear. We study this question in the context of mobile apps, where price variations are common. Specifically, we examine how price promotions in Apple’s App Store affect consumer evaluations measured by subsequent star ratings. Our results suggest that consumers tend to evaluate the product less favorably during price promotions than before the promotions. Further, the negative impact of promotions becomes more pronounced as the discount size increases and is stronger for apps with higher average ratings and/or lower rating variances. A counterfactual simulation suggests that promotions increase downloads for most apps but reduce revenue, and the reputational cost is highest among highly rated apps with few prior ratings. These findings highlight a potential downside of aggressive price promotions for digital products, suggesting that firms must closely monitor online product evaluations to mitigate any negative impacts.
- “Managing Consumer Attention to Diverse Information Sources in Product Diffusion” (2026). Z. Eddie Ning and Zihao Zhou. [Link]
[Abstract]
Before making purchase decisions, consumers use observational learning and may search directly for information from sources with different leanings. This paper develops a theoretical framework in which consumers sequentially decide whether to search for information and which type of information source to search from while observing past prices and purchase decisions of previous consumers. We study how a firm can use dynamic pricing to direct consumers’ attention toward different types of information sources and thereby influence the evolution of the public belief about product quality. The firm may optimally alternate successive consumers’ attention across different types of sources, anchoring the public belief within the search region and preventing herding. Thus, the firm can benefit from a diverse information environment, including sources with unfavorable leanings toward it. Our results remain robust when neutral information sources are available. Our study sheds light on the role of pricing in managing not only consumers’ purchase decisions but also their allocation of attention during the learning process.
- Conference Presentations: EWMES (2025), SICS (2025), Bass FORMS (2025), LQMC (2023), and ISMS (2022)
- “Referral Program Design: Thresholds and Cold Starts” (2026). Zihao Zhou. [Link]
[Abstract]
This paper studies the optimal design of referral programs when customers strategically communicate with multiple friends and trade off referral rewards, concern for their friends’ welfare, and the social cost of making suboptimal recommendations. The firm’s optimal program is generally nonlinear and threshold-based, concentrating rewards on the highest referral outcomes. Such concentration is cost-effective in raising the conversion rate but provides weak marginal rewards at low conversion rates. When customers are highly concerned about making suboptimal recommendations, such programs can suffer from a cold-start problem, in which the conversion rate stagnates at its level without a referral program and remains unresponsive to higher referral rewards. Linear programs are less cost-effective but robust to the cold-start problem. When the firm gradually increases referral rewards within a given class of programs to find the optimal configuration, linear programs provide a wide feedback window that delivers observable improvements in both the conversion rate and the firm’s payoff, whereas threshold-based programs can remain trapped in a feedback desert in which the conversion rate stagnates and the firm’s payoff deteriorates. This implementation contrast highlights a limitation of nonlinear programs and helps explain why firms may favor linear programs despite their lower cost-effectiveness.
- Previously titled “On the Design of Referral Programs”
- Conference Presentation: SICS (2026), Bass FORMS (2026)
- “Flash Pass” (2025). Yuichiro Kamada and Zihao Zhou. [Link] [Online Appendix]
[Abstract]
We consider a model in which an amusement park sells different priority passes to customers whose utilities depend on positions in a queue. A customer’s valuation of a priority pass depends on the distribution of customers buying each priority pass. Hence, other customers’ purchase decisions affect the customer’s valuation, and this externality differentiates our model from the standard screening models. We show that the externality makes the implementation of multi-pass schemes difficult. This issue can persist even when customers have heterogeneous utilities of positions in a queue.
- “Flash Pass with Multiple Customer Types” (2025). Yuichiro Kamada and Zihao Zhou. [Link]
Peer-Reviewed Publications
- “Entrepreneurs and Investors: Funding-Induced Distortions in Lean Startup Product Experiments and Innovation” (2026). K. Sudhir, Onesun Steve Yoo, and Zihao Zhou. Marketing Science, 45(2), 428–448. DOI: 10.1287/mksc.2023.0309 [Eprint] [Online Appendix]
[Abstract]
The lean start-up method (LSM) is an experimentation-driven approach to learning about product-market fit and improving start-up success, yet it overlooks the entrepreneur’s funding needs. This paper models an LSM process in which an entrepreneur tests a product to gauge consumer preferences and seeks funding for a product launch from an investor with incomplete information about the entrepreneur’s type. We show that the investor cannot deduce the entrepreneur’s type from his actions and must rely on the test product’s sales outcome for inference. This reliance can incentivize the entrepreneur to distort his experimentation strategy. Our analysis identifies two distortions in product experimentation relative to LSM without the need for external funding; one prioritizes funding over learning, whereas the other minimizes false positives but discards promising innovations. Consequently, the learning-funding trade-off may induce systemic inefficiency in the LSM ecosystem. Moreover, our model implies that negative investor stereotypes toward certain entrepreneur groups (e.g., women or minorities) may cause them to sacrifice learning to secure funding; hence, when funded, they launch less successful ventures, reinforcing the stereotype. Remarkably, although the inefficiency cost of positive stereotypes is borne by investors, the inefficiency cost of negative stereotypes falls on high-type entrepreneurs from the stereotyped groups.
Research Presentations
The listing is in reverse chronological order.
Conferences
- Sumer Institute in Competitive Strategy | UC Berkeley | 2026
- Annual Bass FORMS Conference | UT Dallas | 2026
- European Winter Meeting of the Econometric Society | University of Cyprus | 2025
- Summer Institute in Competitive Strategy | UC Berkeley | 2025
- Annual Bass FORMS Conference | UT Dallas | 2025
- London Quantitative Marketing Conference | UCL | 2023
- ISMS Marketing Science Conference | UChicago (Online) | 2022
External Seminars and Workshops
2026: HUST; WHU; ZUEL; CUHK; USTC
2021: European Quant Marketing Seminar (Online)
Back to top