> ## Documentation Index
> Fetch the complete documentation index at: https://docs.thunderphone.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Alerts
> Threshold rules on your call metrics — delivered to email, Slack, or your webhooks the moment something slips.
Alert rules watch your organization's call metrics and notify you when
a threshold is crossed — success rate below 80% over the last day,
more than 500 calls in an hour, a test suite regression. Manage them
under **Settings → Organization → Alerts**
(`/dashboard/organization?tab=alerts`). Managing alert rules requires
an **admin** or **owner** role.
## Create a rule
Click **Create rule** and fill in:
| Field | Options |
| --------------------------------- | ----------------------------------------------------------------------------- |
| **Metric** | Success rate · Failure rate · Average score · Call volume · Suite regressions |
| **Condition** | is less than · is at most · is greater than · is at least |
| **Threshold** | The number to compare against |
| **Window (hours)** | How far back the metric is computed — 1 to 720 hours |
| **Email recipients** | Comma-separated, up to 10 addresses |
| **Slack incoming webhook URL** | Optional — posts the alert message to a channel |
| **Cooldown (minutes)** | Minimum quiet time between firings (default 60) |
| **Deliver to developer webhooks** | Also send an `alert.triggered` event to your webhook endpoints |
| **Enable alert rule** | Rules can be saved disabled and switched on later |
Example: *"Success rate **is less than** 80 over the last 24 hours"*
fires when graded calls in the trailing 24-hour window dip below an 80%
success rate.
**Failure rate counts more than "Unsuccessful" calls.** The alert metric
treats every call that didn't reach a successful outcome — including calls
with no conversation at all (immediate hang-ups, failed connections) — as a
failure. The dashboard's **Unsuccessful** tile counts only calls graded
`unsuccessful`. A failure-rate alert can therefore fire even when the
dashboard tile shows a low number; that's expected, not a discrepancy.
## How firing works
Rules are evaluated against the trailing window you configured. When
the condition is met, the rule fires **once** and then stays quiet for
its cooldown period, even if the metric remains in breach — so a bad
hour produces one page, not a page per evaluation. Each firing records
the metric value, threshold, and time.
## Delivery channels
A single rule can notify on all three channels at once:
* **Email** — a message to every listed recipient naming the rule,
metric, value, threshold, and window.
* **Slack** — the same message posted via your Slack incoming webhook
URL.
* **Developer webhooks** — an `alert.triggered` event delivered to your
organization's [webhook endpoints](/webhooks/overview), for wiring
alerts into PagerDuty, on-call tooling, or your own systems.
One channel failing never blocks the others.
## Manage rules
The Alerts tab lists every rule with its enabled/disabled state and
when it last triggered. Toggle rules on and off in place, **Edit** to
change any field, or delete rules you no longer need. Recent alert
activity is shown alongside so you can confirm a rule actually fires
the way you expect.
## Next steps
The dashboards behind the same metrics.
Receive alert.triggered (and every other event) in your own systems.