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

# Incidents Triage: Prioritize Regressions and Opportunities

> Incidents Triage helps you review and prioritize regressions and opportunities detected in your marketing metrics.

Incidents are automatically generated alerts that tell you when something noteworthy has happened in your marketing data. Marcenta runs anomaly detectors across your connected data sources and surfaces two types of findings: **Regressions** (drops or problems) and **Opportunities** (positive shifts you can capitalize on). You don't need to set thresholds manually - the system detects significant metric changes relative to prior periods and assigns a severity level to each finding.

## Navigate to Incidents Triage

Select **Incidents Triage** in the left sidebar. The page opens with **Regressions** selected in the left panel by default, showing you the most recent incidents that need attention.

## Types of incidents

<CardGroup cols={2}>
  <Card title="Regressions" icon="trending-down">
    A metric has declined in a way that warrants attention. For metrics like CTR, conversions, and sessions, a negative delta is a regression. For cost-efficiency metrics like CPC or average position (where lower is better), a positive delta is a regression.
  </Card>

  <Card title="Opportunities" icon="trending-up">
    A metric has improved significantly. These incidents highlight positive momentum you may want to reinforce or replicate across other campaigns or channels.
  </Card>
</CardGroup>

## Severity levels

Every incident carries a severity rating so you can prioritize what to investigate first.

| Severity      | Meaning                                                                            |
| ------------- | ---------------------------------------------------------------------------------- |
| **Very high** | A large, statistically significant change that likely requires immediate attention |
| **High**      | A notable change that warrants prompt investigation                                |
| **Medium**    | A moderate shift worth monitoring                                                  |
| **Low**       | A smaller change that may be within normal variance                                |
| **Very low**  | A minimal shift; informational only                                                |

Use the severity filter in the left sidebar to narrow the list. Select **High** to see only high-severity incidents, or leave it at **All** to see everything for the selected type.

## What information an incident contains

Each incident in the table shows you:

* **Metric** — the specific metric that changed (e.g., Clicks, Click-through rate, Conversions, Bounce rate)
* **Detector** — the anomaly detector that generated the incident
* **Source system** — the connected data source where the change was observed
* **Entity type** — the level at which the anomaly was detected (e.g., global, campaign, channel)
* **Observed at** — when the anomaly was recorded
* **Delta %** — the percentage change compared to the prior period
* **Current value** — the observed metric value at the time of detection
* **Previous value** — the baseline metric value used for comparison

Select any row to open the incident detail view, which includes an AI-generated explanation of the incident, a trend chart, and a breakdown of which campaigns or channels contributed most to the change.

## Manage an incident

From the incident detail view, you can:

* **Acknowledge** the incident to mark it as reviewed
* **Mute** the incident to suppress it without resolving it
* **Resolve** the incident once the underlying issue has been addressed
* **Assign** the incident to yourself, a teammate, or to Marcenta's AI agent for investigation
* **Add a comment** to the activity log to document your findings

<Warning>
  Once an incident is resolved, you cannot move it back to another status. Only resolve an incident when the underlying issue has been fully addressed.
</Warning>

## Run detectors manually

Marcenta's anomaly detectors run on an automated schedule. If you want to check for new incidents immediately - for example, after a major campaign change - select **Run Detectors** at the top of the Incidents Triage page. The page refreshes automatically as results come in over the following seconds.

<Note>
  After triggering a manual detector run, the page polls for new results several times over approximately 20 seconds to capture all completed jobs.
</Note>

## Filter incidents

Use the left sidebar on the Incidents Triage page to filter by:

* **Type** — Regressions or Opportunities
* **Severity** — All, Very High, High, Medium, Low, or Very Low

You can also link directly to a filtered view by adding query parameters to the URL. For example, appending `?type=improvements&severity=high` opens the Opportunities tab filtered to high-severity incidents.

## Page trust signals

* Last reviewed: May 20, 2026
* Maintainer: Marcenta Team