From Billing Insights to Real-Time Streaming Analytics (2018–2025)
Standard metrics that subscription businesses use to measure and compare their performance in the Stripe Dashboard. Formerly known as "Billing Insights" and "MRR". Also used as an input to determine Stripe Capital offers.
Key milestones in the evolution of Billing Analytics
Launched as part of the "Stripe Billing" rebrand (from "Subscriptions"). Sigma queries and visualizations targeting startup & growth-stage companies. Eng DRI: Adam Eagle.
Established as the centralized view for understanding directional business growth, health of customer relationships, and surfacing important metrics merchants might not be monitoring.
Revenue Recovery & Optimization (RRO) team resumed product work on Billing Analytics. Team solicited feature requests and bug reports directly from internal users via go/ask/rro.
Migrated from Scalding-backed Decibel (v7, deprecated) to Pinot-backed Decibel with DataStory for queries. Enabled drill-downs for the first time while maintaining batch processing (24-48h lag).
New Flink-based architecture (ee-glacier) for real-time processing. Parallel pipelines (streaming + batch) ran simultaneously for validation before cutover.
Full cutover to streaming pipeline (April 2025). Metric freshness improved from 24-48 hours to ~10 seconds. All standard Dashboard charts, Sigma exports, and mobile charts migrated. Legacy batch deprecated.
Billing Analytics positioned as part of the broader "Billing Prime" package. Integration with Revenue Analytics organization for a unified analytics experience.
Key initiatives that shaped the product
The single largest technical project — replacing the batch pipeline with Flink-based real-time streaming. Reduced data freshness from days to seconds.
Compass project connecting to Stripe's Reporting Framework. Enables metrics to be consumed by the No Code Data Explorer (RADS SOKR O1KR2).
Allows merchants to compare metrics against industry peers using percentile-based benchmarks with intelligent cohort selection.
Real-time usage and spending data for Usage-Based Billing merchants. Aggregate by meter, customer, dimension, or tenant with rated usage forecasting.
Customer Change Events table for fine-grained metric definitions. Cohort Retention accessible from charts. Template SQL queries for customization.
Streaming transformation integrated with Billing Semantic Events. Data flows downstream to RevRec, Tax Reporting, and other products.
How the technical stack evolved over seven years
| Era | Storage | Processing | Freshness | Drill-down |
|---|---|---|---|---|
| 2018–2022 | Decibel (Scalding) | Daily Spark batch | 24-48 hours | No |
| 2022–2023 | Pinot-backed Decibel | Daily Spark batch | 24-48 hours | No |
| 2023–2024 | Pinot (DataStory) | Daily Spark batch | 24-48 hours | Yes |
| 2024–2025 | Pinot (hybrid) | Flink streaming + batch | ~10 seconds | Yes |
| 2025+ | Pinot (streaming primary) | Flink streaming | ~10 seconds | Yes |
Current streaming architecture (2025)
Common themes from merchant support and internal investigations
Who built and owns Billing Analytics
| Team | Role |
|---|---|
| Revenue Intelligence (formerly RRO) | Primary owning team for Billing Analytics |
| Analytics Core | APIs serving as interface to Stripe's analytics data platform |
| Usage Based Billing | Owns UBB Analytics |
| Reporting Data Interface (RDI) | Infrastructure behind reporting and exports |
| Stream Analytics | Streaming infrastructure (Flink, Kafka) |
| Person | Role |
|---|---|
| Adam Eagle | Original Engineering DRI (2018 launch) |
| Fried Filler | Formerly Billing Analytics; now Team DRI for Usage Based Billing |
| Bill Mei | Software Engineer, Analytics Core team |
Where to find more information