The best tech companies finding product market fit for their Large Language Model (LLM) products start by deeply understanding their users. However, many people in the LLM ecosystem struggle to grasp how and why people use their products, often resulting in a "black box" feeling. To overcome this, they can utilize conversation analytics tools like Context, which helps companies analyze user interactions by labeling transcripts using a taxonomy of categories. By clustering these categories and combining them with per-conversation success reporting metrics, developers can identify key drivers of engagement, product performance issues, and pockets of users showing signs of stronger product market fit. This understanding enables iteration towards achieving product market fit for their LLM products.