A backup that passed and a restore that worked are two different results

Yes 2

Written down, with a date (a log, a screen, a document), and it meets your own plan.

Partly 1

Remembered, but not written down, or the record shows you missed your plan.

No 0

Nobody knows, or it was never done. A blank is a finding.

"None, checked on a date" counts as a record. No plan or target written down? Write your guess first, then measure: the guess is your plan.

Tap No, Partly or Yes for each question. Your score adds up as you go. Nothing you tap is saved or sent.

Five questions about your own setup

1. 1 Date of the last full restore

   Your answer, and where it is written down (with the date) When was the last full restore of your most important application, the whole thing and not a file? Write the date from the log.

   Your answer

   No 0Partly 1Yes 2
   Score 0, 1 or 2This quarterNext quarter

   How to raise it

   Where to find it:

   Your restore log or ticket history for that application. Not the backup job history: a green check on a backup says the job wrote a copy, not that the application came back. No log yet? Time one small test restore into a test folder and write down the minutes and the date. It starts the log.

   To get it to a 2:

   A dated log entry for a full restore, inside the interval your plan sets. No interval written down? Write your guess (once a year, once a quarter) and score against it.

   What this does not prove:

   A restore of one file or one server does not prove the whole application comes back.

2. 2 Time it took

   Your answer, and where it is written down (with the date) How long did that restore take, from the log? Write the hours and minutes next to the target time in the plan (RTO, the recovery time objective).

   Your answer

   No 0Partly 1Yes 2
   Score 0, 1 or 2This quarterNext quarter

   How to raise it

   Where to find it:

   The same log for the start and stop times, and the recovery plan for the target. The clock starts when someone starts working on it, not when data starts copying.

   To get it to a 2:

   A measured time from a log, at or under the written target.

   What this does not prove:

   A restore in a quiet hour does not prove the same time on a busy day, or with several systems down at once.

3. 3 All of it came back

   Your answer, and where it is written down (with the date) What did someone check to show all of it came back, and who said so? (A test order, a report that ties out, and a name.)

   Your answer

   No 0Partly 1Yes 2
   Score 0, 1 or 2This quarterNext quarter

   How to raise it

   Where to find it:

   The test ticket or change record, or ask the application's business owner what they checked.

   To get it to a 2:

   A written list of checks, run by the application's owner, with their name and the date.

   What this does not prove:

   "Restore complete" on a screen does not prove people can use the application.

4. 4 How old the restored data was

   Your answer, and where it is written down (with the date) How old was the copy you restored from? Write its date and time next to the plan's target for data loss (RPO, the recovery point objective).

   Your answer

   No 0Partly 1Yes 2
   Score 0, 1 or 2This quarterNext quarter

   How to raise it

   Where to find it:

   The backup job that made the copy, and the recovery plan for the target.

   To get it to a 2:

   The copy's date and time is in the log, and its age at the time of the test is inside the target.

5. 5 The next full restore

   Your answer, and where it is written down (with the date) Is the next full restore on the calendar, and who runs it?

   Your answer

   No 0Partly 1Yes 2
   Score 0, 1 or 2This quarterNext quarter

   How to raise it

   Where to find it:

   Your team calendar or change schedule.

   To get it to a 2:

   A date on the calendar inside your plan's interval, with a named person.