top of page

Longxiu Tian

TIAN_headshot_2021_2_edited.jpg

​

Welcome! I'm an Assistant Professor of Marketing at the UNC Kenan-Flagler Business School. I received my PhD from the University of Michigan in Marketing and in Scientific Computing.

​

My research is in using causal machine learning and Bayesian econometrics to infer the incrementality of marketing activities, understand how firms should execute lifecycle marketing strategies at scale, and optimize their data strategy.

 

​​I am a co-founder of the Bayesian Computation in Marketing reading group for faculty and PhD students across business schools. Learn more here.

​

Check out my Customer Economics Lab, where students direct an end-to-end customer lifetime value analysis with AI agents, examine the evidence, and make recommendations for a fictional subscription business.

​

ABOUT ME
Research

Research

 

Publications and Working Papers/

Optimizing Price Menus for Duration Discounts: A Subscription Selectivity Field Experiment

Longxiu Tian and Fred M. Feinberg

Marketing Science 39(6), 1181–1198 (2020)

+Show Abstract

Subscription services typically offer duration discounts, rewarding longer commitments with lower per-period costs. The “menu” of contract plan prices must be balanced to encourage potential customers to select into subscription overall and to nudge those that do to more profitable contracts. Because subscription menu prices change infrequently, they are difficult to optimize using historical pricing data alone.


We propose that firms solve this problem via an experiment and a model that jointly accounts for whether to opt-in and, conditionally, which plan to choose. To that end, we conduct a randomized online pricing experiment that orthogonalizes the “elevation” and “steepness” of price menus for a major dating pay site. Users’ opt-in and plan choice decisions are analyzed using a novel model for correlated binary selection and multinomial conditional choice, estimated via Hamiltonian Monte Carlo. Benchmark comparisons suggest that common models of consumer choice may systematically misestimate price substitution patterns, and that a key consideration is the distinctiveness of the opt-out (e.g., non-subscription) option relative to others available.


Our model confirms several anticipated pricing effects (e.g., on the margin, raising prices discourages both opt-in overall and choice of any higher-priced plans), but also some that alternative models fail to capture, most notably that across-the-board pricing increases have a far lower negative impact than standard random-utility models would imply. Joint optimization of the price menu’s component prices suggests the firm has set them too low overall, particularly for its longest-duration plan, and adjusting the price menu accordingly should lead to roughly 10% greater revenue.

/

Probabilistic Machine Learning: New Frontiers for Modeling Consumers and their Choices

Ryan Dew, Nicolas Padilla, Lan E. Luo, Shin Oblander, Asim Ansari, Khaled Boughanmi, Michael Braun, Fred Feinberg, Jia Liu, Thomas Otter, Longxiu Tian, Yixin Wang, and Mingzhang Yin

International Journal of Research in Marketing 43(2), 462–484 (2025)

+Show Abstract

Making sense of massive, individual-level data is challenging: marketing researchers and analysts need flexible models that can accommodate rich patterns of heterogeneity and dynamics, work with and link diverse data types, and scale to modern data sizes. Practitioners also need tools that can quantify uncertainty in models and predictions of consumer behavior to inform optimal decision-making. In this paper, we demonstrate the promise of probabilistic machine learning (PML), which refers to the pairing of probabilistic modeling and machine learning methods, in pushing the frontier of combining flexibility, scalability, interpretability, and uncertainty quantification for building better models of consumers and their choices. Specifically, we overview both PML models and inference methods, and highlight their utility for addressing four common classes of marketing problems: (1) uncovering heterogeneity, (2) flexibly modeling nonlinearities and dynamics, (3) handling high-dimensional and unstructured data, and (4) addressing missingness, often via data fusion. We also discuss promising directions in enriching marketing models, reflecting recent developments in representation learning, causal inference, experimentation and decision-making, and theory-based behavioral modeling.

/

Privacy-Preserving Data Fusion

Longxiu Tian, Dana Turjeman, and Samuel Levy

Marketing Science 45(5), 930–953 (2026)

+Show Abstract

Data fusion combines multiple datasets to make inferences that are more accurate, generalizable, and useful than those made with any single dataset alone. However, data fusion poses a privacy hazard due to the risk of revealing user identities. We propose a privacy preserving data fusion (PPDF) methodology intended to preserve user-level anonymity while allowing for a robust and expressive data fusion process. PPDF is based on variational autoencoders and normalizing flows, together enabling a highly expressive, nonparametric, Bayesian, generative modeling framework, estimated in adherence to differential privacy – the state-of-the-art theory for privacy preservation. PPDF does not require the same users to appear across datasets when learning the joint data generating process and explicitly accounts for missingness in each dataset to correct for sample selection. Moreover, PPDF is model-agnostic: it allows for downstream inferences to be made on the fused data without the analyst needing to specify a discriminative model or likelihood a priori.

 

We undertake a series of simulations to showcase the quality of our proposed methodology. Then, we fuse a large-scale customer satisfaction survey to the customer relationship management (CRM) database from a leading U.S. telecom carrier. The resulting fusion yields the joint distribution between survey satisfaction outcomes and CRM engagement metrics at the customer level, including the likelihood of leaving the company’s services. Highlighting the importance of correcting selection bias, we illustrate the divergence between the observed survey responses vs. the imputed distribution on the customer base. Managerially, we find a negative, nonlinear relationship between satisfaction and future account termination across the telecom carrier’s customers, which can aid in segmentation, targeting, and proactive churn management.

/

The Dynamic Effects of Updates in Crowdfunding: A Provision Point Model

Carlos Siri, Longxiu Tian, and Barry L. Bayus

+Show Abstract

When is the most effective time to post updates during a crowdfunding campaign? Answering this question requires a framework that accounts for the dynamic effects of updates, as well as a provision point mechanism. Crowdfunding platforms like Kickstarter involve an all-or-nothing structure such that no money exchanges hands if the target goal is not met.  Addressing this question raises three interrelated challenges. First, entrepreneurs post updates strategically in response to evolving campaign conditions. Second, the provision point itself is an endogenous decision. Third, the effects of updates are not static but evolve nonlinearly across time. We tackle these challenges by developing a Bayesian provision point model that builds on the insight that entrepreneurs benchmark their decisions against similar past campaigns. This benchmarking provides exogenous variation that allows us to instrument for the non-random timing of updates and correct for endogenous goal-setting. We incorporate Gaussian Process priors to capture smooth, stage-specific dynamics in update effects. We find that updates have a modest and positive effects before the goal is reached, but their effectiveness becomes time-dependent once the goal is surpassed. Our framework also reveals that campaigns setting higher-than-benchmarked goals face weaker pre-goal and stronger post-goal support.  Our research emphasizes the need for entrepreneurs to consider the sequence of updates as these have consequential effects on funding outcomes, and the need to consider the provision point when studying dynamics in crowdfunding.

/

Large N, Small T, Many P: A Causal Tensor Completion Approach for Individualized Promotional Incrementality

Zhongming Jiang, Longxiu Tian, and Fred M. Feinberg

Harvard Business School Working Paper No. 27-021 (2026)

+Show Abstract

A central challenge in contemporary Customer Relationship Management is choosing the optimal intervention for each customer by comparing counterfactual outcomes across a portfolio of promotional policies. This challenge is especially acute in the ubiquitous “large N (many customers), small T (short histories), many P (multiple outcomes)” panel setting, where staggered adoption and pronounced unobserved heterogeneity induce selection on unobservables, undermining traditional causal inference. To this end, we introduce Tensor-Augmented Synthetic Control (TASC), a Bayesian framework that synthesizes synthetic control and matrix completion methods. The core innovation addresses the small T identification problem by leveraging the many P dimension: the method jointly factorizes the full (customer × time × outcome) tensor, using rich cross-sectional variation across outcomes to identify customer-level heterogeneity and enable granular off-policy evaluation even for cold-start customers. An empirical application to an e-commerce startup selling high-end restaurant gift cards suggests four distinct customer response profiles: for ‘Champions’, the optimal policy is a Sitewide-Monthly campaign (+$13.29 in expected incrementality), whereas ‘Dormants’ show no statistically detectable response to any promotion. Together, these results move firms beyond retrospective, on-policy assessment toward prescriptively optimizing personalized promotional strategies, allocating promotional expenditures, and improving profitability.

ChoiceSymposium2_edited.jpg
EDUCATION
EDUCATION
2013-2019

Ann Arbor, MI

UNIVERSITY OF MICHIGAN

Stephen M. Ross School of Business
Ph.D. in Marketing and Scientific Computing

​

Dissertation: Bayesian Nonparametrics for Marketing Response Models (Chair: Fred M. Feinberg)

2011-2012

Cambridge, MA

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

Sloan School of Management
Master in Finance

2004-2008
Evanston, IL

NORTHWESTERN UNIVERSITY

Weinberg College of Arts and Sciences

B.A. in Economics

CONTACT
CONTACT ME

Longxiu Tian

Assistant Professor of Marketing

UNC Kenan-Flagler Business School

​

Office Phone:

919-962-1702

​

Office Address:

300 Kenan Center Dr.

McColl Building 4516

Chapel Hill, NC 27599

​

Email:

longxiu [at] unc.edu

​

LinkedIn:

www.linkedin.com/in/longxiu-tian

Success! Message received.

Copyright © 2026 Longxiu Tian - All Rights Reserved

bottom of page