Notification integrity
A notification you can prove was delivered
Every send writes a row. Failures retry with backoff. An unreviewed service request escalates before it becomes a lost booking.
- Per-channel delivery rows with provider message ids
- Exponential backoff, five attempts
- Terminal failures never retry silently
- Configurable unacknowledged-request escalation
- Weekly failure digest
- Public status page and health endpoint
The request is the revenue
The single most damaging failure in this category of software is an owner not seeing a new client request. The booking goes somewhere else and nobody ever finds out why.
So new service requests always fan out to push, email and SMS, regardless of preferences, and each channel gets its own ledger row before a provider is called.
Retries and a visible failure
A retryable failure backs off at one minute, five, twenty-five, two hours and ten hours. A terminal failure, a dead push subscription, a bad number, stops immediately and raises an admin warning rather than retrying into the void.
The admin header carries a health indicator, and a weekly digest lists everything that failed to deliver so nothing quietly disappears.
Escalation on silence
If a request is still unreviewed after your threshold, two hours by default, the owner gets an escalation SMS and the dashboard shows a persistent banner until someone actually looks at it.