We've improved the median response time of full-text search in Contentful by 35% through a combination of optimizations, including input string normalization, query planning improvements, and schema adjustments. The new implementation uses a SQL function to normalize input strings, allowing for parallelization and pre-evaluation, which reduces query planning time. Additionally, combining text values into a single row per locale improves the performance of full-text searches by leveraging GIN indexes more efficiently. These changes enable editorial teams to quickly locate content using full-text search, tagging, or browsing tag collections, ultimately enhancing productivity and workflow efficiency.