The pattern: weekly review of the page, monthly investigation of trends, alerts on critical metrics. The discipline is the difference between integrations that stay healthy and integrations that surprise the team during incidents.
- Open Data Integration Metrics weekly
Setup, Integrations, Data Integration Metrics. Scan the trends. Note any metric trending in the wrong direction over the past week.
- Check API request volume against the daily limit
Approaching 80 percent of the 24-hour limit is the procurement or optimization trigger. Catch it weeks before exhaustion.
- Investigate trending Bulk API failure rates
Cross-reference with Bulk Data Load Jobs to identify the specific failing job. Fix the underlying integration; do not let the failure rate normalize.
- Monitor CDC and Platform Events delivery rates
Delivery rate below 95 percent signals a subscriber issue. Coordinate with the subscriber team; events queue but the queue overflows eventually.
- Watch Salesforce Connect external object latency
Rising latency signals external system degradation. Coordinate with the external system owner before users complain.
- Build alerts on the critical metrics
API limit at 80 percent, Bulk failure rate above threshold, CDC delivery rate below threshold. Alerts catch degradations between weekly reviews.
- Document the weekly review as ongoing ops
The page only matters if someone reviews it. Calendar entry, named owner, weekly review of 15 minutes. The discipline produces healthy integrations.
Weekly for most orgs; daily for high-volume integration-heavy orgs.
Custom Flow or report-driven alerts on critical metrics (API limit, failure rates).
Bulk API, REST/SOAP API, Streaming, CDC, Platform Events, Salesforce Connect. Pick the relevant subset per org.
24-hour, 7-day, 30-day views. Pick based on the metric's signal cadence.
Documented list of expected integrations with owner and purpose. Required for trend investigation.
- The page does not name specific integrations. Combine with Bulk Data Load Jobs and Connected Apps OAuth Usage to identify the source of trends.
- API limit exhaustion produces immediate integration failures with no recovery until midnight UTC. Catch the trend weeks before; do not wait for exhaustion.
- CDC and Platform Events queue events when delivery fails but the queue is bounded. Subscriber outages beyond the queue lifetime lose events permanently.
- Salesforce Connect latency depends on external systems Salesforce does not control. Coordinate root cause with external owners; do not blame Salesforce when the external is slow.
- Without integration ownership, trend investigation stalls. Maintain an integration inventory with owner per integration; trends without owners become orphans.