An example for the sign "a dozen measures": one page that the operations team goes through once a month in half an hour. The company and the numbers are made up. The page prints on one A4 sheet.
| Measure | October | September | Target | 6 months | Status | What is behind it |
|---|---|---|---|---|---|---|
| What the business saw | ||||||
| 1. Downtime of key serviceshours, planned windows included | 7.4 h | 11.2 h | ≤ 8 h | on target | Of these, 3.0 h are planned windows. The longest outage was 2.1 h, email | |
| 2. A client reported the outage firstshare of priority incidents | 22% | 12% | ≤ 10% | over the limit | 2 of 9: both in the client portal, the availability check does not see it | |
| 3. Time to restoreaverage for priority incidents | 1 h 40 min | 2 h 10 min | ≤ 2 h | on target | 7 of 9 closed by the on-call engineer with an instruction, no expert | |
| Repeats and debts | ||||||
| 4. Repeat outagessame object and cause within 90 days | 13% | 16% | ≤ 10% | attention | 4 of 31. Three are one switch in office 2, a problem record already exists | |
| 5. Problem recordsopen / overdue | 7 / 1 | 6 / 0 | 0 overdue | attention | Overdue: power supply swap in the storage system, waiting for the supplier | |
| 6. Temporary workarounds older than a quartercount in use | 1 | 2 | 0 | on target | One left: manual restart of the export to accounting, due in December | |
| Load on the team | ||||||
| 7. Alert noiseshare that needed no action | 58% | 64% | ≤ 50% | attention | Half of the noise is CPU load thresholds on 4 servers | |
| 8. Night callstotal / by the written rule | 3 / 3 | 5 / 4 | all by rule | on target | Not a single call "just in case" | |
| Changes and data | ||||||
| 9. Failed changesrollback or outage in the first 24 hours | 9% | 11% | ≤ 10% | on target | 2 of 22. Both rollbacks on time by plan, no downtime for users | |
| 10. Complete recordsincidents with time, object and outcome | 96% | 94% | ≥ 95% | on target | The numbers for this page took 10 minutes, with no manual cleanup | |