Beyond the Dashboard: Building Trustworthy Data with Observability at PAYBACK
summary
PAYBACK shows concrete steps to make data trustworthy: dbt unit tests in CI/CD that catch logic regressions, row-level dbt data tests and regex checks for validity, BigQuery DataPlex profiling with DataPlex sampling profiling to limit scan cost, and Datadog dashboards and alerts that surface failed Airflow jobs, query latency, and GCP costs. A central dashboard already routes composer failures to team channels, and an end‑to‑end testing plan that includes partners is under active development.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:44Trustworthy data matters now
- 6:14Data product observability pillars
- 10:07Resource monitoring for data infra
- 11:56Execution monitoring beyond pipelines
- 13:22Data quality monitoring essentials
- 17:57Checks at rest vs in-flight validation
- 20:27PAYBACK observability principles
- 24:24DBT testing and unit tests
- 29:48Auto DQ scans with DataPlex
- 32:19DataDog dashboards
- 34:27Key takeaways and future plans
- 36:12Q&A
related talks
21mJourney from Broken Dashboards to O11y Maturity
35mApplying Data Quality Monitoring
47mObservability - Holistic Quality across Software Systems
30mBeyond the Dashboard: Build End-to-End O11y
29mData Observability: From 1788 To 2032
33mAnalytics Data Driven Testing
42mObservability Is Not Analytics!
12mUnlocking the Secrets of Client Data