Overview
Amplitude emphasizes event quality and user identity for reliable behavioral cohorts. Plan the rollout with a strict taxonomy, environment keys, and ingestion governance to prevent noisy data.
Implementation Phases
Discovery & Access
- Gather Amplitude API keys/secret keys per project environment and confirm data center choice (US/EU).
- Define identity strategy: device_id vs user_id rules, and whether to enable merged users.
- Agree on the taxonomy for events and properties; decide whether to enforce via Governance features.
Instrumentation Build
- Initialize the Browser SDK with the correct API key and set default tracking options (sessions, page views) intentionally.
- Register user properties (plan, role, experiment variants) after identify/setUserProperties calls.
- Implement server ingestion via HTTP V2 API or Batch API for orders and backend actions with consistent user_id/device_id values.
- Configure ingestion filters or transformations to drop PII or unexpected properties.
QA & Launch
- Use Amplitude's Event Stream/Monitor to validate payloads, property casing, and device_id/user_id behavior.
- Check sessionization and attribution (UTM/referrer) across SPA route changes and auth redirects.
- Confirm that EU endpoints are used if the project is EU-resident and that ad blockers are mitigated via proxy if needed.
Deployment Artifacts
- Taxonomy/tracking plan mirrored in Amplitude Govern with owners and descriptions.
- API keys, secret keys, and ingestion endpoint references for each environment.
- Identity strategy doc covering device_id, user_id, and merged user settings.
Linked Runbooks
- Install or Embed the Tag or SDK
- Event Tracking
- Data Layer Setup
- Cross-Domain Tracking
- Server-Side vs Client-Side
Change Log & Owners
- List owners for Govern approvals and who can change tracking plan statuses.
- Track open questions about default tracking settings (sessions/pageviews) and their impact on quotas.