Incidents
Correlate incidents from PagerDuty and OpsGenie with deployments and PRs — and track MTTR as a DORA metric.
Overview
Incidents are the most expensive events in an engineering organization — both in direct cost (engineer time, customer impact) and in what they reveal about system reliability. Koalr's Incidents page integrates with PagerDuty and OpsGenie to bring incident data into your engineering metrics context, showing you not just when incidents happen but why — and which deployments preceded them.
Integrations
Incidents require at least one of:
- PagerDuty — See PagerDuty integration
- OpsGenie — See OpsGenie integration
Once connected, Koalr syncs incidents in real time via webhook and backfills the past 90 days.
OpsGenie sunset notice: Atlassian is sunsetting OpsGenie on April 5, 2027. Koalr supports both platforms and can help you migrate — see OpsGenie migration guide for details.
MTTR — Mean Time to Recovery
MTTR is one of the four DORA metrics. Koalr calculates it as:
MTTR = average(incident resolved_at − incident triggered_at)For services with multiple triggered alerts per incident, Koalr deduplicates using the earliest trigger and latest resolution.
| DORA Tier | MTTR |
|---|---|
| Elite | < 1 hour |
| High | < 1 day |
| Medium | < 1 week |
| Low | > 1 week |
Dashboard sections
Organization MTTR summary
- Current MTTR (median and P75)
- DORA tier classification
- Incident count in period
- MTTD (Mean Time to Detect) — trigger time minus deploy time for deployment-correlated incidents
Incident timeline
Chronological list of all incidents in the selected period, showing:
- Incident title and severity
- Service/integration
- Duration (trigger → resolution)
- Whether it was correlated with a deployment
- Post-mortem status (if post-mortem linked in PagerDuty/OpsGenie)
MTTR trend chart
12-week line chart of median MTTR. Also shows P75 to capture tail incidents. This is your DORA Lead Time for Recovery trend.
Deployment correlation
This is Koalr's most powerful incident feature: automatically correlating incidents with the deployments that preceded them.
For each incident, Koalr looks back 2 hours (configurable) and identifies any deployments that occurred in that window. It then shows:
- Which PR was deployed (with risk score at time of deploy)
- Which team deployed it
- Whether the pre-deploy risk score was High or Critical (early warning ignored?)
This closes the feedback loop: teams can see which deploy risk scores translated into actual incidents.
PR Autopsy
For every incident, Koalr automatically runs a PR Autopsy: it finds all pull requests merged into the affected service's repositories in the 2-hour window before the incident triggered, scores each PR's likelihood of being the culprit, and generates an AI-written root-cause hypothesis.
The autopsy shows:
- Each merged PR with its deploy risk score at time of merge
- A "culprit score" ranking (risk score × recency × change size)
- A one-paragraph AI hypothesis identifying the most likely contributing change
Click any incident in the timeline → PR Autopsy tab to view.
See PR Autopsy for full documentation.
Service reliability table
Per-service metrics:
- Incident count
- MTTR
- MTTD
- Change Failure Rate (incidents per deployment)
- Last incident date
Team comparison
Which teams have the best/worst MTTR and incident frequency. Helps identify which teams need process support vs. infrastructure support.
On-call burden
How many hours each engineer spent in active incidents (from PagerDuty escalation data). Surfaces on-call imbalance — a burnout signal.
Change Failure Rate (CFR)
CFR is computed alongside incident data as:
CFR = incidents correlated with a deployment / total deploymentsThis is the DORA "Change Failure Rate" metric. Koalr displays it on both the Incidents page and the DORA Metrics page.
| DORA Tier | CFR |
|---|---|
| Elite | 0–5% |
| High | 5–15% |
| Medium | 15–30% |
| Low | > 30% |
Post-mortem tracking
If your PagerDuty or OpsGenie incidents are linked to post-mortem documents (Notion, Confluence, Google Docs), Koalr surfaces:
- Whether a post-mortem exists for each incident
- Post-mortem completion rate (incidents with completed post-mortem / total incidents)
- Action item tracking (if structured post-mortem format is used)
Configuring incidents
Navigate to Settings → Deployments to configure the deploy correlation window (default: 2 hours before incident trigger). Service-to-repo mappings and severity filters are configured in your connected PagerDuty or OpsGenie account and synced automatically. Post-mortem reminders require the Slack integration to be connected.
Relationship to other metrics
- DORA Metrics — MTTR and CFR are two of the four DORA metrics (DORA Metrics)
- Deploy Risk — incidents are the outcome that deploy risk signals predict; correlation closes the feedback loop (Deploy Risk)
- Burnout Risk — on-call incident hours are an input to burnout risk scoring (Burnout Risk)
- SPACE Metrics — incident rate feeds the Performance dimension (SPACE Metrics)
- PR Autopsy — automatically surfaces the most likely culprit PR for each incident, reducing diagnosis time (PR Autopsy)
E2E Test Coverage
Track code coverage across repos with Codecov and SonarCloud — and surface coverage delta as a deploy risk signal.
Weekly Risk Briefing
AI-generated weekly risk narrative delivered to engineering leaders every Monday — high-risk PRs, coverage declines, recent incidents, and reviewer expertise gaps summarized in plain English.