Data Governance and Integrations: Taxonomy, Sources, and Destinations
What the exam tests
This domain covers how data enters and leaves Amplitude. Sources send events in (SDKs, HTTP API, CSV upload, cloud storage, warehouse). Destinations export data out (warehouses, analytics tools, webhooks). Know the tracking plan concept: a documented schema of approved events and properties that keeps taxonomy consistent. Understand data volume controls, event blocking, and merge/identity rules that govern governance.
Key mechanics
- Instrumentation: client-side SDKs vs. server-side HTTP API; the API is preferred for trusted, backend events.
- Warehouse integrations (e.g., Snowflake, BigQuery) for importing and exporting.
- Data governance tools: tracking plans, blocked events, and property validation reduce junk data.
- Identity resolution merges anonymous and identified users; misconfigured IDs fragment users.
Traps to watch
- Assuming an SDK is required when the HTTP API suffices for server events.
- Forgetting that blocking an event stops ingestion but does not retroactively delete historical data.
- Confusing import (warehouse to Amplitude) with export (Amplitude to warehouse).
- Overlooking that inconsistent user IDs split one person into multiple users.
Rule of thumb
Treat the tracking plan as the contract: every event must be named, typed, and owned. Worked example: a team wants backend purchase events - send them via the HTTP API server-side, register them in the tracking plan, and validate property types so downstream funnels stay reliable.