Reducing alert fatigue

Alert fatigue kills monitoring programs. Intello gives you severity, cause grouping, scopes, and channel allow-lists—this page shows how to combine them.

Start with routing rules

Build a table before touching the UI:

Risk	Channels	Scope
Breaking + `payments` domain	PagerDuty + Slack `#payments-oncall`	`domain=payments`, `min_severity=breaking`
Risky / safe	Email digest daily	`api`, `min_severity=safe` (digest mode via external mail rules)
Informational	In-app only	`notify_in_app` on provider prefs; no chat channels

Prefer group acknowledgements

When a schema change fans out, ack the cause_key group once, not each endpoint row. This keeps Slack threads honest about remaining work.

Tune provider noise separately

Catalog vendors may ship noisy specs. Use:

spec_filters / domain_filters on provider alert preferences
is_muted on irrelevant provider spec slices

…before muting the entire vendor.

Avoid duplicate dependencies

Two dependencies pointing at the same logical spec doubles events. Consolidate.

Use health states intentionally

A dependency stuck in paused health is a signal—fix the URL or delete the integration. Silent disablement accumulates blind spots.

Review quarterly

Archive dependencies for retired integrations.
Delete experimental alert prefs.
Confirm PagerDuty rules still map to real rotations.