Grips Research

Optimizing Product Matching in E-commerce This research explores product matching and its vital role in e-commerce to match products across different platforms. We propose a multi-stage, deep learning-based system and evaluate its performance based on the precision-recall trade-off. Read Article Shopping Missions in Online Grocery Shopping This study introduces a novel mission-based model for segmenting consumers in the online grocery market, leveraging extensive transaction data from a leading U.S. supermarket chain. Utilizing BERTopic modeling, we analyze shopping basket compositions to identify distinct consumer shopping missions. Read Article Evaluating The Impact of Privacy Regulation on E-Commerce Firms Assembling novel datasets on online advertiser spending, performance, and revenue, we quantify the economic effects of Apple's App Tracking Transparency (ATT) privacy policy on e-commerce firms. Our paper extends prior literature by uncovering the effect of ATT on not only advertising effectiveness but also changes in firm-side strategy to mitigate the effects of ATT and the resulting net impact on overall firm revenue. Read Article Sanctions, Sales, and Stigma: A Tale on the Performance of International Brands in Russia The full-scale Russian invasion of Ukraine in February 2022 posed a dilemma for many international brands: withdraw from the Russian market for reputational concerns or stay to make profits. We study this trade-off using novel, detailed in- formation on customer transactions of 95 global brands from 1,774 webshops that act as intermediaries. Read Article Ukraine - Firms through the War 2.0 This report focuses on the Ukraine war and its impact on the following: (i) conflict matters; (ii) firm resilience and adaptation at limit; and (iii) direct government support remains low. Read Article ROAS Management Strategically managing Return on Ad Spend (ROAS) can enhance profitability. We analyze 500+ advertisers and show that a simple ROAS-based budget reallocation can increase sales by 5%-25% without additional ad expenditure. Neural product embeddings for (new) product revenue prediction We show that product embeddings outperform characteristics-based revenue predictions models and propose a method to infer embedded product vectors for new unknown products. We conduct a meta study on 150+ online retailers to validate the method. Beyond Rankings: Exploring the Impact of SERP Features on Organic Click-through Rates Supplemental SERP features, including knowledge graphs and video reels, significantly impact CTR prediction. Our analysis of 100,000+ keywords suggests a 20% increase in prediction accuracy using these features. Read Article Interpretable Deep Learning for Forecasting Online Advertising Costs: Insights from the Competitive Bidding Landscape Cost Per Click (CPC) prediction is essential for marketing strategy planning. Our study of 12,000 advertisers shows that Dynamic Time Warping can identify indirect competitors, improving long-term prediction. For example, data from flour sales can improve predictions for kitchen appliance marketing. Read Article Russia’s e-commerce trade in the aftermath of the 2022 invasion: Evidence from high-frequency data Despite many brands leaving Russia following the 2022 Ukraine invasion, retailers remain. Our analysis of parallel imports and 100+ brands' cross-border e-commerce sales helps estimate the effects of public pressure on companies to cease Russian trade. Read Article Online Advertising Revenue Forecasting: An Interpretable Deep Learning Approach Most advertising performance prediction research focuses on the advertiser's perspective. Our study, covering 200+ news websites' ad revenue, highlights the benefits of training a temporal fusion transformer on all available data. Read Article Teaching an old fox new tricks: How sustained e-commerce sales lift from pandemic lockdowns depends on age COVID-19 lockdowns spiked online sales, but customers under 50 returned to baseline shopping habits two months later. The 50+ generation, having adapted to online shopping, maintain high sales levels post-lockdown.