The industry is struggling with integrating Large Language Models (LLMs) into products, proving Return on Investment (ROI), and understanding user goals. Building a great LLM product requires iterative process of prototyping, testing, and improving over time, driven by deep understanding of users' needs and goals. The challenge lies in scaling to analyze conversational data and identifying user intent, which is typically solved using text analytics that labels conversation transcripts on two dimensions: meaning and success. This provides visibility into how and why people are using the product, where user needs are being met, and where improvements can be made.