Monitor Operations with Business KPIs

KPI dashboards are most useful when reviewed in a consistent sequence during incidents or operational reviews.

Read KPIs During an Incident

  1. Select the dashboard for the affected business capability.

  2. Set the dashboard time window to the suspected incident period.

  3. Check the request or activity rate to confirm whether relevant business activity occurred and whether demand changed.

  4. Check the error or failure rate to identify whether failures increased during the same period.

  5. Check mean duration and latency percentiles to determine whether average or tail behavior shifted.

  6. Review any KPI-specific panels for the affected business capability.

  7. Compare KPI movement with releases, traffic changes, network events, synthetic checks, support tickets, and downstream system signals.

  8. Preserve the relevant dashboard time window and screenshots when sharing incident evidence.

Choose a Time Window

Start with a narrow time window when the approximate event time is known. Expand the window to establish a baseline before and after the event.

Some business KPI metric families are calculated over rolling windows. Very short dashboard windows can make trends appear sparse. Use wider windows for baseline comparison and shorter windows for active incident tracking.

For normal operations, choose a refresh interval that matches operational needs and observability platform capacity. For active incidents, use a shorter refresh interval once the collector and datasource are confirmed healthy. Do not configure the collector to scrape /kpi/metrics more frequently than once every 20 seconds.

Correlate KPI Signals

Correlate business KPI movement with evidence from systems outside the service boundary. Useful comparison points include:

  • Availability checks

  • Release and configuration changes

  • Traffic management changes

  • Support ticket volume

  • Downstream dependency health

  • Integration logs

  • Alert and notification history

Endpoint Errors

Status Operational meaning

401 Unauthorized

The request did not include a valid access token.

403 Forbidden

The access token was not accepted for the KPI Service environment or scope.

503 Service Unavailable

The KPI source is temporarily unavailable.

504 Gateway Timeout

The KPI source did not respond within the service timeout.

500 Internal Server Error

The service could not render a metrics response.

Observability tooling should alert on failed scrapes separately from business KPI alerts. A failed scrape means the KPI signal is unavailable; it does not indicate the health of the monitored business capability.

Troubleshoot Scrapes

Symptom Checks

The token endpoint returns invalid_scope

Confirm the client has the KPI Service scope assigned in the token issuance policy and that the request uses the fully qualified scope.

The token endpoint returns unauthorized_client

Confirm the confidential client is active and has the Client Credentials grant enabled.

The KPI Service returns 401

Confirm the bearer token is current and was issued by the expected identity domain.

The KPI Service returns 403

Confirm the token audience and scope match the KPI Service resource application and environment.

The collector receives HTML or an empty response

Confirm the scrape URL is /kpi/metrics, the bearer token is being sent, and the request is not routed through an incorrect proxy.

Responsibilities

Area Your organization controls Oracle manages

Endpoint consumption

Metrics collector configuration, scrape interval, datasource setup, and observability platform integration.

Secured Business KPI Service endpoint, access controls, and service operation.

Identity client and secrets

Confidential client lifecycle, client secret storage, token retrieval, and token refresh.

KPI Service resource application, required scope, and onboarding details.

Metric storage

Retention, storage location, backup, access to observability data, and long-term reporting.

KPI source data handling inside the Oracle-managed service boundary.

Dashboards

Imported dashboard copies, layout changes, panel thresholds, annotations, folders, and sharing.

Published Grafana dashboard templates and template availability.

Alerts

Business alert thresholds, notification routing, escalation policy, and alert suppression rules.

Stable KPI definitions and metric naming.

KPI definitions

Business interpretation, desired thresholds, and requests for additional business KPIs.

KPI aggregation logic, business capability mapping, exposure safety, and onboarding of new KPI definitions.

Incident analysis

Correlation with systems outside the service boundary, operations timelines, and incident records.

Shared KPI evidence for exposed Oracle Health Insurance business capabilities.

Capture Support Information

When raising a support question about the Business KPI Service, include the following:

  • Tenant and environment context

  • Affected Business KPI or dashboard

  • Approximate time window, including time zone

  • Affected metric family or panel

  • Endpoint status code if scraping failed

  • Issue type (empty data, failed scrape, unexpected value, token issuance, authorization, or dashboard import behavior)

  • Relevant correlation points, such as releases, traffic changes, or incidents outside the service boundary

Do not include credentials, access tokens, client secrets, or private endpoint secrets in support material.