When product decisions at large online marketplaces affect millions of buyers and sellers daily, guessing wrong costs real money. Nihar V. Patel has spent his career making those decisions measurably better. As a product data scientist at a leading global marketplace platform, he has identified substantial business growth opportunities through rigorous experimentation and causal analysis. His work translates messy marketplace data into clear strategic direction, helping product teams understand not just what happens when they change features, but why it happens and who benefits most.
The challenge Patel tackles daily centres on a deceptively simple question: when a platform changes something for sellers, how does it ripple through to buyers, and back again? Traditional A/B testing struggles in two-sided marketplaces because the two groups constantly influence each other. Changes in seller pricing shift buyer behaviour, and that movement circles back, altering how sellers respond. Standard statistical methods often break down under these conditions, leading to biased results that can mislead teams.
Turning Experiments Into Reliable Answers
Patel’s technical contributions span seller onboarding, product listing optimisation, and emerging AI-powered commerce features across web and mobile applications. His method relies on causal inference frameworks that account for the interdependence between marketplace participants. Rather than assume groups behave independently, he designs experiments and analyses that acknowledge their connections. Through randomised A/B experiments and quasi-experimental methods such as difference-in-differences analysis, he measures how product changes affect outcomes across the entire ecosystem.
His most successful interventions have driven meaningful increases in both seller and buyer engagement across multiple product areas. Beyond headline metrics, Patel builds measurement systems for products with multiple downstream effects, enabling teams to distinguish true causal impact from mere correlation. His evaluation of classification models rigorously tests accuracy and performance trade-offs, verifying that model-driven decisions translate into measurable business value before teams ship them.
Earlier in his career, at a major online deals-and-promotions marketplace, Patel worked on the seller marketing and advertising side of the platform. There, his deep dives into buyer purchase behaviour and advertiser performance uncovered growth levers that led to significant gains in ad performance and return on ad spend. The pattern holds across both roles: understand user behaviour at a granular level, translate insights into testable hypotheses, and guide product teams towards data-backed decisions.
Building Methods Others Can Use
Beyond his day-to-day product work, Patel has authored six research papers addressing specific technical challenges in marketplace experimentation. These papers, several of which are slated for publication in respected indexed journals, offer practical frameworks that other practitioners can apply.
His propensity score matching framework addresses selection bias in marketplaces where shoppers and sellers self-select and algorithms optimise in real time. The method provides concrete diagnostics that platforms can directly build into experimentation tools, turning causal inference from a statistical exercise into an operational process for quasi-experiments.
His synthetic control work adapts econometric methods to the context of ride-hailing, where cities differ in infrastructure, regulation, and user adoption. The framework handles staggered rollouts and time-varying policy intensity, providing platforms and regulators with credible impact estimates without assuming parallel trends. His research on heterogeneous treatment effects examines how driver incentives vary across neighbourhoods, time periods, and market saturation levels, enabling more efficient and equitable resource allocation.
Regarding algorithmic fairness, Patel developed an eight-stage lifecycle framework to mitigate bias in two-sided marketplaces. The framework treats exposure and opportunity as scarce resources requiring fair allocation over time, addressing feedback loops that can amplify small initial biases into major systemic problems.
A Path From Petroleum To Platforms
Patel’s route to data science began in petroleum engineering, where analytical thinking and uncertainty modelling form the core of the skill set. Reservoirs behave unpredictably, requiring engineers to build models from incomplete data and test predictions against reality. Those same skills transfer directly to marketplace analytics, where user behaviour resists easy categorisation and teams must act despite uncertainty.
His postgraduate work in engineering management, emphasising product and data science, formalised the transition. Recognition came early: he won his university’s Husky Startup Challenge and later took first place in an internal product development competition at his company. Outside work, he built Caldris, an AI-driven app that generates personalised workout plans and routines.
The through-line in Patel’s career is clear: take complex systems generating noisy data, apply rigorous analytical methods, and extract actionable intelligence. Whether measuring reservoir performance or marketplace health, the discipline remains constant. His work helps ensure that when platforms make changes affecting millions of users, those changes rest on evidence rather than assumptions.
