> ## Documentation Index
> Fetch the complete documentation index at: https://docs.wittify.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Alerts
> Catch trouble before it compounds.
**Alerts** is the page that watches the numbers that matter and tells you when something is unusual.
## Page anatomy
| Region | What you see |
| ------------------- | ------------------------------------------------------------------------------------ |
| Filter bar | Date and hour controls, see [Overview](/en/Systems/Contact-Center-QA/Overview-Page). |
| Aggregate KPI strip | Four small cards with feed-level totals. |
| Tab bar | **Anomaly feed** and **Rules** pills. |
| Feed or rules table | Body of the active tab. |
| Event detail drawer | Slides in when you click a row in the feed. |
A small mock-data notice strip can also appear above the KPI strip when the project has just been connected and rules have not yet seen real data, with copy *Rules are evaluated live against your analytics data for the selected window, compared against the equal-length prior window. Rules you create persist in your browser until the server-side evaluator lands and takes over email, webhook, and in-app delivery.*
## Aggregate KPIs
Four cards reactive to the active filter window.
| KPI | Notes |
| --------------------- | --------------------------------------------------------------------------------------------------- |
| **Open anomalies** | Anomalies that are not yet acknowledged or resolved. |
| **High and critical** | Subset filtered to **High** and **Critical** severity. |
| **Rules enabled** | Count of rules currently turned on. |
| **Last detected** | Timestamp of the most recent anomaly. Reads **No anomalies in range** when there are none in scope. |
## Anomaly feed tab
The default tab. A time-ordered feed of anomalies for the active filter window.
| Element per row | What it shows |
| ----------------- | --------------------------------------------------------------------------------- |
| **Severity** chip | One of **Low**, **Medium**, **High**, **Critical**, in green, amber, orange, red. |
| **Triggered by** | The rule that produced this anomaly. |
| **Detected at** | A timestamp formatted **YYYY-MM-DD HH:MM:SS**. |
| **Observed** | The metric's actual value in the alert's window. |
| **Baseline** | What the metric usually is over the equal-length prior window. |
| **Change** | Delta between observed and baseline. |
| **Status** chip | **Open**, **Acknowledged**, or **Resolved**. |
| Click | Opens the event detail drawer (see below). |
When the feed is empty, the empty state reads *No anomalies in this range. Everything looks steady against the rolling baseline. Widen the date filter to scan a longer window.*
## Rules tab
A project-scoped library of alert rules.
| Column | Notes |
| ------------- | ----------------------------------------------------------------------- |
| **Rule name** | Plain-language description. |
| **Metric** | The thing being watched (negative sentiment share, call volume, etc.). |
| **Condition** | The operator and threshold. |
| **Window** | The evaluation window. |
| **Severity** | The chip color. |
| **Delivery** | Channels: **In-app**, **Email**, **Webhook**, or any combination. |
| **Enabled** | A switch to turn the rule on or off. Off rules stop firing immediately. |
| **Updated** | Timestamp of the last edit. |
A **Create rule** button on the card header opens the rule dialog. Each row also carries an inline **Edit** action to modify an existing rule and a **Delete** action with a destructive confirmation dialog.
### Default seed rules
Six rules ship by default, five enabled.
| Rule | Condition | Severity | Default state |
| ---------------------------------------- | -------------- | ------------------------- | ------------- |
| Negative sentiment share at or above 10% | `>= 0.10` | High (in-app and email) | Enabled |
| Negative sentiment spike | rises by 15% | Medium | Enabled |
| Call volume shift | rises by 20% | Medium | Enabled |
| Low call volume | drops by 30% | Medium (in-app and email) | Enabled |
| Handle time above 90 seconds | `> 90` seconds | Low | Enabled |
| Silence ratio at or above 30% | `>= 0.30` | Medium | Disabled |
## Creating a rule
Click **Create rule** in the Rules tab. A small dialog opens.
| Step | What you do |
| ------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Name** | Type a name your team can spot in the feed. Optional **Description** for context. |
| **Metric** | Pick from **Negative sentiment share**, **Call volume**, **Avg handle time**, **Escalation rate**, **Silence ratio**, **Dead-air events per call**. |
| **Operator** | One of **is above**, **is at or above**, **is below**, **is at or below**, **rises by**, **drops by**. |
| **Threshold** | A number within the metric's range. The dialog rejects out-of-range values inline with a small red error. |
| **Window** | One of **Last 15 minutes**, **Last hour**, **Last 4 hours**, **Last 24 hours**. |
| **Severity** | **Low**, **Medium**, **High**, or **Critical**. |
| **Delivery** | At least one of **In-app**, **Email**, **Webhook**. Save is blocked when no channel is selected. |
The dialog saves the rule to the project. New anomalies that match this rule appear in the feed on the next evaluation pass.
## Event detail drawer
Click any row in the Anomaly feed to open a side drawer. The drawer also opens automatically if you arrive via a deep link (the URL carries the event id).
| Section | What it shows |
| -------------------------------- | --------------------------------------------------------------------------------------------------------------------- |
| Header | Severity chip, rule name, detected timestamp. |
| Detail block | Observed value, baseline, delta, the metric being watched. |
| **Open Sentiment Analysis** link | Routes to [Sentiment Analysis](/en/Systems/Contact-Center-QA/Sentiment-Analysis) with the alert's window pre-applied. |
| **Open Overview** link | Routes to [Overview](/en/Systems/Contact-Center-QA/Overview-Page) with the alert's window pre-applied. |
| Calls table | The actual calls captured in this alert's window. |
| Search | Filter calls by filename, session id, or transcription. |
| **Acknowledge** button | Flips the alert's status to **Acknowledged**. |
| **Resolve** button | Flips the status to **Resolved** when the issue is fixed. |
| **Download CSV** | Exports the calls in this window. |
The drawer fetches calls for the alert's own window, not the page's filter range. So a 1-day alert always shows the right calls even when your filter bar is set to **Last 30 days**.
## Topbar bell
The global notifications bell in the dashboard topbar swaps to the CCQA alert feed whenever you are on a CCQA route. The bell trigger gains a small red count badge whenever there are unread anomalies, and the dropdown body lists the most recent few. The dropdown header has a **Mark all read** action.
There is one notification bell across the entire app. CCQA does not add a second bell, the existing bell extends to show alerts when you are on a CCQA route.
## Common questions
The rule may genuinely never have triggered (good news). The other common cause is the metric is not yet computed in the active filter window, change to a wider date range and check again.
Yes. Open the Rules tab, click the **Delete** action on the rule's row, confirm the destructive dialog. The rule disappears immediately. To bring it back, recreate it with **Create rule**.
Email and webhook delivery rely on a server-side dispatcher that is being rolled out. Today only the in-app bell reflects new events. The Email and Webhook channels are saved on the rule, the dispatcher will read them when it goes live.
The rule's window is independent. A rule has its own window (15 minutes, 1 hour, 4 hours, 24 hours) and is evaluated continuously. The page's filter only changes which anomalies you see in the feed, not when rules fire.
Yes. The drawer scopes calls to the alert's window, not the page filter, and adds the alert metadata at the top. Use the **Open Sentiment Analysis** link to jump to the project-wide view with the alert's window pre-applied.
Not yet. Each project has its own rules library, and you create or delete them through the dialog. Bulk import is on the roadmap.
The bell counts unread (Open) anomalies. Once you acknowledge or resolve an event, it stops counting toward the bell badge.
Mark them **Resolved** in the drawer. They stay in the feed history but no longer count as Open. There is no automatic auto-resolve today.
## Where to go next
Set the targets these rules monitor.
Drill into the calls behind a sentiment alert.
Silence tab feeds the dead-air events metric.
Pre-empt a low-call-volume alert.