Statsig's Knowledge Graph: Connecting code, experiments, and metrics

Statsig is developing a knowledge graph to enhance team learning and improve response times when handling product changes and alerts. As products grow more complex, the learning curve for understanding how changes affect outcomes becomes steeper, often hindering teams' ability to respond quickly. The knowledge graph aims to bridge this gap by making explicit the connections between product components, metrics, and outcomes, which traditionally exist only in the mental models of experienced engineers. By integrating these relationships into a machine-readable format, both human teams and AI systems can more effectively trace the causes of changes and implement targeted solutions, reducing the reliance on speculative fixes. This infrastructure not only facilitates faster debugging and clearer experimentation but also empowers AI systems to function with the contextual understanding of seasoned engineers, thereby optimizing the cycle of shipping, measuring, learning, and iterating.