In this session, we share the journey of delivering “Cost per Feature” information that bridges the gap between raw cloud spend and product value. We will detail a methodology that combines cloud billing data with distributed tracing to attribute costs at the feature level—even across complex, asynchronous backend service chains.
Key takeaways for attendees:
The Methodology: How to use Open Telemetry trace span duration as a proxy for compute resource consumption and precise token counts for AI service attribution.
Connecting the Stack: Strategies for linking frontend user interactions to backend infrastructure costs using a unified Telemetry SDK and trace context propagation.
Navigating Accuracy & Sampling: Managing the trade-offs between data volume and precision, including how to statistically adjust for trace sampling to provide reliable directional insights.
Driving Business Value: Practical applications for feature-level data, from evaluating the ROI of premium features and informing pricing models to incentivizing engineering efficiency.