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

# Volume alerts

> Email notifications when a tool's response volume drops or spikes abnormally relative to its history

A **volume alert** watches how many responses a tool receives and
notifies by email when the current period falls outside its historical
behavior. When you create it, you choose which of the two conditions you
want to watch:

* **Response drop**: the current volume falls well below the historical
  average.
* **Response spike**: the current volume rises well above the historical
  average.

Volume alerts follow woku's general alert model: condition, recipients,
frequency, and dispatch history. That model is described in
[Alert rules](/docs/en/configuracion/reglas-de-alerta).

## What is measured

Each alert measures a tool's response volume: Wokus, NPS, Forms, CSAT,
or CES. You can measure all of the tool's instruments together, or a
specific instrument: a particular woku, NPS, form, CSAT, or CES.

<Note>
  The count considers production responses only. Test data does not affect
  the evaluation.
</Note>

## Comparison periods

The period defines which history the current volume is compared against:

* **24 hours**: the current day against the daily average of the last
  4 weeks. Requires at least 7 days of history.
* **7 days**: the current week against the weekly average of the last
  4 weeks. Requires at least 2 weeks of history.
* **30 days**: the current month against the monthly average of the last
  3 months. Requires at least 2 months of history.

## How it decides

The alert fires only when the current period meets **both** conditions:

1. It crosses the configured **percentage threshold**. For a drop:
   `actual < promedio * (1 - umbral/100)`. For a spike:
   `actual > promedio * (1 + umbral/100)`.
2. It is **2 or more standard deviations** away from the historical
   average (z-score).

If the history is perfectly flat (zero standard deviation), the
percentage threshold alone is enough.

The threshold is configured between 1% and 100%. The form suggests 40%
for drops and 80% for spikes as a starting point.

For example, with a 7-day period, an average of 200 weekly responses
over the last 4 weeks, and a drop threshold of 40%, the alert fires when
the current week records fewer than 120 responses and that decline is
also 2 or more standard deviations away from the average.

This dual condition makes the alert deterministic and explainable: the
notification always corresponds to a verifiable fact, such as "volume
dropped more than 40% relative to the average of the last few weeks",
and not to an opaque criterion.

## Insufficient data

If the tool or the instrument has not yet accumulated the minimum
history for the chosen period, the alert stays in an **insufficient
data** state and does not fire. The alert detail shows the message "Not
enough data yet" until there is enough history to compare.

## Email notification

When the condition is met, the alert sends an email to the configured
recipients. The frequency controls how often the notification is
repeated:

* **As soon as it happens**.
* **Once a day while it keeps occurring** (default option).
* **Once a week**.

## Evaluation and history

woku evaluates alerts automatically once a day. In addition, in the
alert detail you can press **Evaluate now** to force an immediate
evaluation. Every notification sent is recorded in the alert's
**Dispatch history**.