Research Interest
​
1. Introduction​
My name is Pukar Koirala, and I recently completed my Master’s in Marketing Analytics & Insights. My interest in research began when I noticed how quickly AI-generated answers and ranking algorithms were reshaping the way people search for information, evaluate options, and make decisions. This experience sparked my curiosity about how digital platforms influence human behavior and how algorithms quietly structure the choices people see.
​
2. Research Background
During my graduate studies, I completed two major research projects that shaped my academic direction.
In Generative AI and the Transformation of Modern Marketing Practices, I examined how AI-powered search summaries reduce click-through rates, change user attention, and shift the distribution of online visibility. Analyzing this structural change helped me understand the growing influence of AI as an information gatekeeper.
​
In my second study, How AI Shapes Consumer Decisions: An XAI Study of MBA Program Choice, I used machine learning and SHAP explainability to analyze how variables—such as financial readiness, career motivation, perceived ROI, and engagement—drive high-stakes educational decisions. This project introduced me to the importance of transparent AI systems that not only predict outcomes but also reveal the reasoning behind them.
These experiences strengthened my interest in decision science, digital platforms, and interpretable AI.
​
3. Current Research Interests
​
• AI-Driven Search, Ranking Systems & Digital Behavior Shifts
I am interested in how AI-generated answers, search overviews, and ranking algorithms reshape user behavior and information visibility. My focus is on understanding how trust forms, how attention is allocated, and how decision patterns change when AI curates the first—and often the only—answer users see. I aim to examine how these systems influence consumer choice, reduce exploration, and affect fairness and competition across digital markets.
​
• Explainable Machine Learning & Human Decision Processes
I am focused on using interpretable machine learning to understand why people make certain decisions under uncertainty. Building on my MBA-choice study, I want to explore how psychological readiness, financial constraints, information overload, and AI-generated cues interact to shape real decisions. My goal is to create transparent, human-centered models that reveal the reasoning behind predictions and help institutions make better, data-informed decisions.
RESEARCH - 1
How AI Shapes Consumer Decisions: An XAI Study of MBA Program Choice
​
Author: Pukar Koirala
Date: November 2025
​
Description
This study examines how prospective MBA applicants make decisions, focusing on financial readiness, career motivation, ROI expectations, and emotional confidence. Using machine learning (Logistic Regression, Random Forest, XGBoost) and Explainable AI (SHAP), the research reveals why individuals choose—or avoid—an MBA program.
A key finding is that AI-driven interactions (chatbots, AI summaries, recommendation engines) strongly increase awareness, engagement, and intention to apply. Behavioral and psychological factors matter far more than demographics, showing that modern education decisions depend on perceived value, reduced uncertainty, and AI-mediated guidance.
This project provides universities and marketers with a clear framework for understanding student behavior and designing data-driven recruitment strategies.
Skills
Used AI Interaction Research • XAI (SHAP) • Machine Learning Models • SEM FoundationsConsumer Behavior Analysis • Survey Design • Predictive Modeling • Data Interpretation
RESEARCH - 2
Generative AI and the Transformation of Modern Marketing Practices
​
Author: Pukar Koirala
Date: October 2025
Description
This research project investigates the structural and behavioral impact of generative AI on modern marketing systems. Through a comprehensive analysis of large language models, generative visual technologies, and AI-driven analytic tools, the study examines how algorithmic generation reshapes core marketing workflows, including ideation, content production, forecasting, optimization, and strategic decision-making. A central focus of the research is the emergence of AI-powered search interfaces and the rise of Answer Engine Optimization (AEO).
The study evaluates how AI Overviews, multi-turn conversational search, and summary-based information delivery alter organic visibility, consumer search pathways, and competitive dynamics. To address these changes, the project introduces a new set of measurement frameworks—AI Visibility Rate (AIVR), Answer Inclusion Share (AIS), and Reference Depth Score (RDS)—designed to quantify brand presence within AI-generated responses.
The findings highlight several future research directions related to consumer trust in AI-mediated information, the tension between automation and creativity, market power concentration among high-authority domains, and ethical considerations surrounding AI-driven persuasion. Overall, the project offers a structured foundation for understanding how AI is reconfiguring marketing ecosystems and outlines implications for practitioners, scholars, and policymakers navigating this evolving landscape.
Skills Used:
Research Design, Quantitative Analysis, AI Marketing, Algorithmic Search Behavior, SEO/AEO Modeling, Data Interpretation, Digital Strategy
RESEARCH - 3
Behavior of Small-Budget Advertisers in Repeated Ad Auctions
​
Author: Pukar Koirala

Year: 2025

Project Overview
This project analyzes how small-budget advertisers behave in repeated digital advertising auctions, focusing on how limited budgets, volatile pacing, and ranking uncertainty reshape bidding and learning dynamics. The study connects practical advertising behavior with core concepts in optimization, online learning, and mechanism design within the DRO domain.
​
Research Focus
The project investigates three key questions:
1. Budget Constraints: How limited budgets shape learning, bidding, and error sensitivity.
2. Pacing Stability: How pacing algorithms over- or under-spend in volatile auction environments.
3. Ranking Volatility: How small shifts in ad position impact CPC, impression share, and outcomes.
Method & Approach
* Modeled advertisers using dynamic optimization and bandit-style learning under budget uncertainty.* Analyzed pacing behavior across varied budget levels.
* Simulated ranking shifts to measure their effect on efficiency and spending.​Key Insights* A 1–2 position drop raised CPC by ~21% for small advertisers.
* Unstable pacing caused faster budget depletion in volatile auctions.* Sparse feedback created high learning volatility compared to high-budget advertisers.
* Budget-limited agents deviated from classical auction equilibrium predictions.
Research Contribution
The study highlights how real-world constraints generate different optimization behaviors than standard models assume.
Findings offer implications for:
* Better pacing algorithms
* Fairer allocation mechanisms
* More robust learning and feedback systems for constrained advertisers
Skills & Tools
Auction Analytics • Optimization • Online Learning • Simulation • Google Ads • Research Writing
Curriculum Vitae
For a complete overview of my academic background, research experience, technical skills, and professional journey, please refer to my full CV below. It highlights my work across marketing analytics, AI-driven research, and data-focused projects, along with my publications and ongoing studies.