GenAI-Powered Customer Sentiment Analysis Unlocks $8M in Savings
Gen AI, Retail Analytics, Product Insights
Background
A global apparel manufacturer with a large omnichannel presence received thousands of online product reviews but lacked a scalable way to extract actionable insights. Attribute-level feedback (e.g., fit, fabric) was buried in unstructured text, return reasons were unclear, and manual review processes were slow and inconsistent—limiting timely business response.
Impact
- $8M annual savings in COGS through reduced returns and better product fit
- 25–27% decrease in return rates, particularly for size-related issues
- Improved CTR and customer satisfaction via enhanced product pages with summarized reviews
Solution
Affine deployed a Generative AI–based review summarization and sentiment engine, enabling faster, structured insight extraction from customer feedback:
- Data Pipeline: Consolidated review and attribute data across SKUs using Databricks
- GenAI Layer: Leveraged Azure OpenAI (GPT-4o) with custom prompt engineering to extract sentiment by product attributes
- Dashboard: Delivered topic- and sentiment-based summaries of top reviews for business consumption
- Human-in-the-loop validation ensured quality and reliability
- Fully integrated into monthly product analytics for design and merchandising teams
Impact
- $8M annual savings in COGS through reduced returns and better product fit
- 25–27% decrease in return rates, particularly for size-related issues
- Improved CTR and customer satisfaction via enhanced product pages with summarized reviews
Recommended Case Studies
Talk to the experts!
