What is an analytics monitor?

Kinds of rule

• Threshold — the value crosses a fixed line: error rate above 2%, orders per hour below 5.
• Period change — the value moves sharply against a previous period: conversions down 30% against the same hour last week.
• Forecast band — the value falls outside the range predicted from its history.
• Anomaly — the value is unusual given its seasonal pattern, without you choosing a number.

Thresholds are easy to understand but need tuning; anomaly rules adapt to seasonality but are harder to explain when they fire.

Avoiding alert fatigue

An alert that fires every day gets ignored, and then the real one is missed. Habits that help:

• Alert on outcomes (orders, sign-ups, errors), not on every metric.
• Use a coarser granularity for noisy metrics: daily instead of hourly on small sites.
• Set severity, and send only critical alerts to channels that wake people up.
• Before enabling a rule, check how often it would have fired in the past.

Monitors in TapCub

TapCub monitors can watch any saved metric or built-in measure with eight kinds of condition: threshold, period change, forecast band, anomaly, funnel drop, frustration signals, segment size and no data. Each has an hourly, daily, weekly or monthly granularity and an info, warning or critical severity. A dry run shows how many times the rule would have fired in the last 30 days before you turn it on.

Alerts go to the project's notification channels: email, WeCom, DingTalk, Feishu or a webhook, with ready-made webhook templates for Slack and generic JSON.

Common mistakes

• Hourly thresholds on low traffic. Random dips will fire constantly.
• No "no data" rule. The most dangerous failure is tracking that stops entirely.
• Alerts without an owner. Every alert should name who acts on it.

Related guides: Web Analytics Best Practices and Product Health Metrics. See also set up traffic alerts.