Analytics Alerts

An alert watches one metric on your site and notifies you when it crosses a threshold — across the whole site, on a page path, or for a specific event. TinyAnalytics delivers the alert through an organization-wide notification destination — email, Slack, Discord, or webhook — and sends an all-clear when the metric returns to normal.

​

Start from a template

The create dialog starts with ten templates that prefill the metric, condition, and alert name:

• Traffic spike +50%
• Traffic drop −50%
• Page traffic
• Event volume
• Event users
• Revenue −20%
• Exit-rate spike +10pts
• AI traffic surge — AI share of traffic up 50% versus the previous day
• AI traffic drop — AI share of traffic down 50% versus the previous day
• Blank

After choosing one, complete the three sections: What to watch, When to notify, and Where to notify. The last section uses destinations shared by your organization. Open Advanced to change the window, check cadence, cooldown, or saved segment. When you edit an alert with non-default advanced settings, that section opens automatically.

​

What can you alert on?

Pick the metric to watch and the window to measure it over — the last hour or the last day:

​

Scope an alert to a page or event

For a page-specific alert, add a Page path and choose is, contains, or starts with. The path scope is available for non-event metrics and can also be used with event metrics. An Exit rate alert always watches exactly one page: its Exit page field is required, matches one exact path, and stands in for the page-path scope (segments don’t apply to exit-rate alerts — the watched page always travels with the alert itself). An AI share of traffic alert is site-wide and stays silent when the window holds fewer than 10 sessions — too small a sample to call a share; it neither fires nor reports 0%. It accepts only session-level scopes (a channel, country, entry page, or a saved segment of that kind); event-level filters such as a page path are rejected, so the Page path field is hidden for this metric. Unlike exit rate, it has no settling delay. For Event count or Event users, enter the event name and optionally add up to five property filters. Property filters use exact key/value matches, all joined with AND. For example, an event-users alert can watch distinct users who fired checkout_failed with both plan = pro and country = DE.

​

Set the condition

An alert fires when the metric crosses a bound you define, in one of two ways:

• Absolute — the value goes above an upper bound or below a lower bound (for example, sessions in the last hour drop below 50).
• Relative — the value changes by more than a percentage versus the previous equal window (for example, pageviews jump more than 40% hour over hour).

You can preview the current value and whether it would breach as you type the thresholds, and send a test notification to confirm your destinations work before you rely on the alert.

​

How often can an alert fire?

• TinyAnalytics checks each alert every minute. It fires once when the metric first breaches, then holds until the metric recovers — it won’t re-fire every minute.
• When the metric returns within bounds, TinyAnalytics sends a recovery all-clear.
• A cooldown (12 hours by default) keeps a flapping metric from spamming you: a repeat breach inside the cooldown is suppressed. Recovery notices are never suppressed.

A fire notification includes the site, watched metric, window, breach reason, and the top pages and referrers at the time, so you can start investigating from the message itself. Email alerts include a direct View alert link. Recovery messages clearly mark the alert as resolved and do not repeat stale spike context.

​

Frequently asked questions

​

Related