Sedi
PhD Candidate in Quantitative Marketing
Warwick Business School, University of Warwick
Job Market 2026–27
Department of Marketing, Warwick Business School
University of Warwick, Scarman Rd, Coventry, UK CV4 7AL
I am a PhD candidate in Quantitative Marketing at Warwick Business School, and I am on the academic job market for 2026–27. My research sits at the intersection of econometrics, particularly causal inference, and computational methods. I study how digital platform design shapes consumer behaviour, market outcomes, and welfare.
My substantive research examines the consequences of platform interventions for consumers and other participants. In commercial settings, I combine quasi-experimental analysis of large-scale transaction data with meta-analytic evidence to study how discount design influences consumer decisions, sellers, and market outcomes. In digital health, I investigate how different types of nudges shape engagement, access, and consumer behaviour.
My methodological research explores how large language models can support survey-based research. I develop approaches that use LLM-generated responses alongside human data to improve inference, particularly when collecting primary data is costly or limited.
Ph.D., Marketing (Candidate)
Warwick Business School, University of Warwick
M.Res., Business & Management (Marketing)
Warwick Business School, University of Warwick — Distinction
MSc, Business Administration
Amirkabir University of Technology — GPA: 18.3/20
BSc, Computer Software Engineering
Sharif University of Technology — GPA: 16.3/20
Drawing on longitudinal transaction data from a large food-delivery platform, this paper examines how planned versus immediate discounts shape consumer purchasing behaviour. It focuses on how discount timing interacts with consumers’ tendencies to explore new restaurants, and how these differences relate to spending, restaurant choice, and responses to promotions.
Using four LLM families (ChatGPT, Gemini, DeepSeek, Grok) across six published marketing experiments, we develop a framework that improves LLM-human correlations from r ≈ 0.08 to r ≈ 0.40 via multi-model ensembling, covariate recalibration, and response-style adjustment integrated through Recalibrated Prediction-Powered Inference (RePPI). The pipeline reduces standard errors by approximately 16% on average, corresponding to an effective labeled-sample-size increase of roughly 24%.
A meta-analysis of effect sizes from academic papers (2017–2026) decomposing promotional timing into signalling (immediate vs. planned) and reward delivery (instant vs. delayed). Instant rewards significantly outperform delayed ones; however, this disadvantage is nearly eliminated when consumers receive advance notice, enabling anticipatory mental accounting.
When a seller funds a discount on an online marketplace, whose demand does it take? Using transaction-level data from a food delivery platform, I build a competitor network from shared pre-period customers and estimate how seller-funded coupons affect rival sellers. Geographic definitions of competition show no rival losses, but revealed competition does: losses concentrate on a narrow set of rivals that share customers with the promoting seller, and are steepest among those selling the same type of food. How competitors are measured determines whether promotional spillovers are visible at all.
Examines how different types of digital nudges influence consumer engagement and behaviour on a digital health platform, exploring variation in responses across users and intervention types.
Develops a Bayesian approach to using LLM-generated responses alongside evidence from related experiments to help assess new research manipulations.
Evaluates whether LLM predictions can supplement a small human survey (N ≈ 1,200 Scottish residents via Prolific) to produce valid constituency-level estimates of voting intention for the 2026 Scottish Parliament election, combining human and synthetic responses through Prediction-Powered Inference.
“Beyond Single LLMs: Using Model Ensembles to Improve Synthetic Data in Marketing Experiments”
INFORMS Marketing Science (ISMS). Oral presentation, 2026.
“Beyond the Bargain: Deciphering the Complexities of Food Choices”
European Marketing Academy (EMAC). Poster.
Marketing Research Brownbag
Warwick Business School.
Marketing and Strategy Analytics
Graduate Teaching Assistant, MSc — Warwick Business School. Rating: 4.3/5 (2025).
Principles of Marketing
Graduate Teaching Assistant, Undergraduate — Warwick Business School.
Managerial Decision Making
Graduate Teaching Assistant, MBA — Amirkabir University of Technology.
Computer Programming
Teaching Assistant, BSc — Sharif University of Technology.
Teaching Excellence Certificate
Recognition for contribution to Warwick Business School (WBS).
Product Manager — Snapp!
Developed the in-house A/B testing and experimentation platform; applied econometric methods to evaluate causal effects of platform interventions.
Product Analyst — Tapsell
Analysed large-scale advertising data to optimise business operations; built dashboards and data models supporting data-driven decision-making.