This document discusses the challenges of search relevancy and provides examples of techniques used to improve search results, including: 1) It explains concepts like precision and recall that are used to evaluate search results and the difficulty of determining what constitutes a "good" result. 2) It provides examples of how search can go wrong when queries are ambiguous and shows different types of search queries people commonly use. 3) It describes techniques like DSSM and C-DSSM that use deep learning to build robust representations of queries and documents to improve relevancy.