Define which record decides
Name the authoritative booking record and the calendar record it should correspond to. Use synthetic or approved fixture data only. Record each ID, the calendar scope, the operator, and what staff are allowed to do when the records disagree. Do not assume that a visible calendar event means the booking system accepted a change.
Google's incremental synchronization guidance describes storing a sync token, retrieving changed and deleted entries, and doing a full resync when the service invalidates a token. Those are API behaviors to check against a real integration; they do not prove that your booking workflow has mapped the right record or policy.
Run one bounded change and removal
- Create one approved fixture booking and record its authoritative ID and calendar counterpart.
- Save the checkpoint the integration says it will use for the next synchronization.
- Change one harmless field and observe whether the corresponding calendar record is identified and updated according to the approved rule.
- Remove or cancel the fixture through the approved path and observe the deleted-entry or removal handling.
- Record the full-resync decision: who performs it, what local state is replaced, and how the booking record is reconciled afterward.
The local evidence fixture rejects an incomplete declaration:
node --test sites/howtox.com/evidence/P131/calendar-sync-receipt.test.mjs
It does not call an API, send an appointment message, prove delivery timing, or validate a real provider configuration.
Stop when the mismatch has no owner
Do not rely on the boundary if the authoritative booking record, calendar counterpart, stored checkpoint, changed/deleted observation, full-resync procedure, or reconciliation owner is unknown.
For a timed-process boundary, see Test a Scheduled Job Before Launch. For a separate message-delivery exercise, see Test an Application Notification Before Launch.
Does a passing fixture prove the live schedule is correct?
No. It records one narrow exercise. Real permissions, calendar scope, conflict policy, operational timing, and production data handling still need review in the approved environment.