VoidhashDocs
DocumentationAPI reference
Dashboard
Get started
Introduction
Quickstart
Setup
Setup overview
Create a project
Environments
Products and perks
Paywall locations
CLI
Paywall components
Components
Develop a component
Editor panels
Component definition
Publishing
Installing and capabilities
SDKs
Swift
Kotlin
React Native
Web
Store setup
Connect the App Store
Connect Google Play
Connect Stripe
Record payments from other sources
API libraries
Node.js
PHP
Rust
Go
Resources
MCP server
Alerts
Ad measurement
Meta ads
Migrate in observer mode
Delete a person's data

Alerts

Hear about failures and missing events without watching a chart

An alert watches your project's events and tells your team's owners and admins, by inbox and email, when something looks broken. Create them under Analytics > Alerts in Studio, or with the CLI, the API or an agent over MCP.

Alerts read production events and production purchases. They are checked once an hour, at a quarter past, over windows that closed on the hour.

What an alert can watch

KindFires whenExample
Too many of an eventAn event happens at least so many times, to at least so many people.purchase_failed 5 times in an hour
Too many failuresThe share of attempts that failed is clearly above your limit.upload_failed against upload_completed
An event dropsAn event falls by your limit against the same window in the weeks before.$pageview down 50% over 24 hours

People affected. A count needs at least 2 different people by default, so one device retrying does not fire it. Set it to 1 when a single person matters, such as someone who paid and has no access.

Attempts. For failures, send one event when an attempt fails and another only when it worked. An attempt is either event, so if everything fails the alert reads 100%.

When an alert stays silent

On a small project a quiet window is normal, so an alert says when it cannot tell instead of guessing.

  • Not enough volume. Too few events in this window to judge. At 2 purchases a day, no purchases in an hour means nothing, so choose a longer window. A failure alert also needs at least 5 failures before it fires.
  • Not enough history. A drop is compared with earlier weeks, and this event was first seen too recently to have two of them. Only time fixes this; the alert starts judging on its own.

How often you hear from it

You get one notice per problem, not one per hour. An alert stays open while the problem lasts, and clears itself after two clean checks and at least six hours, so something flapping stays one notice. People can mute these notices in their notification settings.

Changing what an alert watches, or turning it off, starts it over and closes any open notice within the hour. An alert can only tell you what your events say: a broken release that stops sending events reads as a drop, and an event you never send reads as nothing at all.

PreviousMCP serverNextAd measurement

On this page

What an alert can watchWhen an alert stays silentHow often you hear from it