> ## Documentation Index
> Fetch the complete documentation index at: https://docs.obsy.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Reading RCA results

> Understand each section of the RCA report and how to act on it.

## Report sections

### Summary

A concise paragraph answering: "What happened?" Read this first to quickly orient yourself and share context with teammates who weren't on the initial response.

### Root cause

The single most likely technical explanation for the incident. Examples:

> "A misconfigured circuit breaker in the `payment-service` deployment (v2.4.1, deployed at 14:32 UTC) caused downstream timeouts to cascade into the `checkout-service`, resulting in 503 errors for 8% of users."

This is what you should fix. If you disagree with the root cause, re-run the analysis after adding more context (e.g. linking the correct deployment via Change Intelligence).

### Contributing factors

Secondary issues that didn't cause the incident but made it worse, harder to detect, or harder to fix. These are candidates for follow-up improvements.

Examples:

* "No circuit breaker timeout on the `checkout-service` calling `payment-service`"
* "Error rate alert threshold was set too high, delaying detection by 12 minutes"
* "Deployment lacked a canary stage; 100% of traffic hit the bad version immediately"

### Impact

What was affected, for how long, and how severely:

* Affected services
* Error rate increase (e.g. "5xx errors increased from 0.1% to 8.3%")
* Duration from first alert to resolution
* Whether the status page was updated

### Recommendations

Specific, actionable items ranked by priority. Each recommendation maps to either the root cause fix or a contributing factor improvement.

* **Fix the root cause** — concrete steps to resolve the issue
* **Prevent recurrence** — structural changes (circuit breakers, better timeouts, canary deployments)
* **Improve detection** — alert tuning, runbook links, SLO adjustments

### Impact score

A 0–1 numeric score combining severity (50% weight), signal type (30%), and duration (20%). Use it to prioritize postmortems — scores above 0.7 warrant a full blameless postmortem.

| Score range | Suggested action                                           |
| ----------- | ---------------------------------------------------------- |
| 0.8–1.0     | Full postmortem, executive summary                         |
| 0.5–0.79    | Internal postmortem, action items tracked                  |
| 0.3–0.49    | Short incident review, recommendations logged              |
| \< 0.3      | Document in incident timeline; no formal postmortem needed |

## Acting on the report

1. **Share the summary** in your incident Slack channel or status page update.
2. **Assign the root cause fix** as an incident action item.
3. **Schedule the contributing factor improvements** in your sprint backlog.
4. **Create a postmortem** if the impact score warrants it — click **Create postmortem** from the RCA detail page.