This document discusses recommendations at Zillow. It outlines Zillow's goal of helping users discover homes that match their needs through personalized recommendations. It describes the different areas recommendations could be used, such as email, search, and individual home pages. It then covers the key components of building recommendations, including datasets, metrics, modeling techniques like collaborative filtering and content-based modeling, infrastructure for offline and online testing, and evaluation metrics. Finally it discusses Zillow's data sources and machine learning pipelines for feature engineering, training, prediction, and evaluation of recommendation models.