Alert Noise Reduction Audit

Cut pager noise and keep the alerts that matter.

Added October 15, 2026. Written with AI assistance and checked for clarity and placeholders; not reviewed by a lawyer, accountant, clinician or other professional. How these prompts are made

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Audit my alerts for noise. Export of the last 30 days of alerts with name, count, time to acknowledge, and whether action was taken: [ALERT_EXPORT]. On-call rotation size: [ON_CALL_SIZE]. Service tiers: [SERVICE_TIERS].

Rank alerts by volume and by percentage that led to no action. For each of the top fifteen produce a decision: delete, downgrade to a ticket, add a delay or duration, aggregate, fix the cause, or keep. Explain your reasoning briefly and state the evidence from the export. Calculate how many pages per on-call shift the changes would remove, in a before-and-after table. Define a rule that every page must be urgent, actionable, and require a human, and test each kept alert against the rule. Propose a review routine: a monthly alert review with a standing agenda, an owner for each alert, and a policy that alerts without an owner are removed after 60 days.

Include a short section on what to do with alerts that fire rarely but matter a great deal, so that the clean-up does not remove the alert that would have caught a serious event. Ask me for the cause of the last three major incidents and check that each would still have paged someone after your proposed changes. Present the final alert list sorted by owner.

Add a worked example of one decision so the reasoning is visible: an alert named high-cpu-web that fired 212 times in 30 days, was acknowledged in a median of 11 minutes, and led to action 4 times. Show the percentage with no action, the decision (replace with a duration condition and an alert on latency, which is what users feel), the pages per shift removed, and the check that it would not have hidden any of the last three major incidents. Then add a section that groups the remaining alerts by owner and states, for each owner, how many pages per week they will own after the changes. Provide a ninety-day follow-up plan with three measurements (pages per shift, percentage of pages with action, after-hours pages) and a target for each that I can edit.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

Top alerts by volume (30 days)
CPU > 80% on web: 412 pages, 3% led to action. Decision: change to sustained 15 minutes; page only if latency also rises.
Disk 70%: 120, 0% action. Decision: ticket, not page.
5xx > 1%: 38, 71% action. Keep.
Estimated pages per shift before: 31; after: 9.
Rule: a page must be urgent, actionable and need a human.
Ownership: any alert without an owner is removed after 60 days.
Safeguard: rare but important alerts (backup failure) stay.

• alerts