Information seeking behavior, as measured by Google searches, and information sharing behavior, as measured by the language in Twitter posts, give complementary insights into the mental and physical well-being of communities. Language use in Facebook and Twitter has been widely used to assess well-being, but much less attention has been given to search patterns. We train statistical models on both Google search volume in different categories and Twitter language frequency across different topics to predict life satisfaction, emotional well-being, and a variety of health outcomes and related behavioral risk factors across 208 Designated Market Areas (DMAs) in the United States. Google search volume generally outperforms models trained on Twitter topics in predicting a wide variety of outcomes, from life satisfaction to excessive alcohol and tobacco use. Twitter language only outperforms Google search in predicting community-level stress. Our findings suggest that Google searches provide insights into lifestyle and community culture and may prove useful to monitor well-being.