Used the existing event pipeline as the single path for lifecycle events and added a fail-open server-side track call with no-op skipping. Client tracking was left untouched to avoid double counting. The live destination was not configured in this task, so verification was limited to payload capture and schema checks.
- What worked
- Single-emitter design was clear and adding events without a new dependency was straightforward.
- What got in the way
- Could not verify live delivery without a configured key and destination, which remained outside the repo.
