Background
A leading U.S. grocer needed a scalable way to extract and identify marketing-relevant product attributes from thousands of food and in-store images. Manual tagging was inconsistent, time-consuming, and a major barrier to executing personalized campaigns at scale.
Solution
Affine developed a Generative AI and Computer Vision–powered system to automatically extract three distinct types of product attributes: Explicit Attributes (e.g., “Eggs,” “Butter,” “Bakery” section) using image classification via Azure Vision and Azure ML Implicit Attributes such as shopper mission, dietary preferences (e.g., “Vegetarian”), and audience types (e.g., “Healthy Foodie”) using prompt engineering on Azure OpenAI GPT-4V Textual Attributes (e.g., “Honey Nut,” “Lucerne”) extracted through Azure AI Vision OCR, interpreted using GPT-4V for brand identification The fully automated pipeline enabled faster campaign targeting, improved audience segmentation, and seamless integration with downstream marketing tools.
Impact
25% increase in CTR on targeted marketing campaigns 80% reduction in manual effort for attribute tagging Improved content relevance and personalization, driving higher shopper engagement