Customers often search for very specific product features that are difficult to discover from only the supplier-provided product information. For example, how do you rank products for customers searching for an “easy to clean rug” or a “comfortable to sit on couch.” To help e-commerce search engines evolve and embrace this challenge, we need to look beyond the product catalog to improve query understanding and product ranking. In this session we will show how to address this problem by leveraging customer feedback datasets such as product reviews. We will take a deep dive into some natural language processing techniques such as opinion mining, sentiment analysis to discover, refine and rank useful insights from reviews, apply sentiment models to label mined data and then show how to integrate this information into Solr in order to surface the right products in search results. We will also discuss how this work improved conversion rate for some obscure customer searches on Wayfair.com.