Defining the number

Backup success is successful jobs divided by scheduled jobs across the period. A job that did not run counts as a failure, because a job that did not run protected nothing. A job that ran with warnings counts as a success only if a restore from it has been tested; otherwise it is unknown, and unknown is a data gap the page should name.

Ninety-eight percent across four hundred jobs is eight failures. If the eight are one server, the client has a system with no protection for eight nights. That is the finding, not the ninety-eight.

The sentence that lands

Backup success ran at 98.4 percent; every miss was the finance file server, which has not completed a job since the storage change on the ninth. The number sets the context, the noun makes it real, the date makes it checkable.

What to ask for

The decision is rarely about backups. It is about the thing blocking them: a storage quota, a service account, a device the client retired without telling anyone. Bring the cause, propose the fix, ask for the approval or the budget, set the date.

  • Approve the fix for the failing system, with a date.
  • Confirm which systems are in scope for backup at all.
  • Agree when the next restore test happens and who witnesses it.

Keeping it comparable

Backup tools report differently, and the temptation each quarter is to show whichever screen looks best. Export one number per period under one written rule and keep the two rows. A brief built from those rows, with the source row beside the claim, is what a client can trust across quarters. That is the discipline Sourcerow enforces; the restore test is still on you.

A worked example

A logistics client with six servers and two NAS units runs 420 scheduled jobs a quarter. This quarter 413 succeeded, which is 98.3 percent. Six of the seven failures are the same warehouse NAS, which lost its network share when the client's electrician replaced a switch in July. The seventh is a one-off on the mail archive that succeeded the following night.

The finding on the page is one line: backup success held at 98.3 percent; the warehouse NAS has had no successful job since the switch change on 12 July and is the only unprotected system. The decision requested is to approve the two-hour visit to reconnect and re-seed it, dated before month end. Presented that way, the client hears one exposed system and one fix, not a percentage they are supposed to be pleased about.