Anomaly Detection

Pro Plan

Anomaly detection automatically flags unusual patterns in your analytics. It runs as part of AI Insights, so significant changes surface alongside your other insights instead of waiting for you to spot them.

What Is Anomaly Detection?

Each time Zenovay generates insights for your site, it compares your recent daily metrics against a 30-day baseline and flags days that fall well outside the normal range:

Monitors	Detects
Visitors	Spikes and drops
Bounce rate	Unusual increases
Visitor value score	Sudden drops in average quality

These anomalies are scored as info, warning, or critical depending on how far the day deviates from the baseline.

How It Works

Statistical Analysis

Zenovay builds an understanding of "normal" from your own history:

Your typical day this month:
- Visitors: 1,000-1,200
- Bounce rate: 42-48%

Today:
- Visitors: 580 ← ANOMALY (well below the usual range)

A day is flagged when a metric moves more than two standard deviations away from its 30-day average. Moves beyond three standard deviations are marked critical.

Baseline Period

When you start:

Zenovay needs at least a few days of daily data before it can detect anything (it uses a rolling 30-day baseline).
As more days accumulate, the baseline becomes more representative.
With a full month of history, detection reflects your real day-to-day variation.

If a site has fewer than three days of historical data, no anomalies are reported yet.

Continuous Updates

Because the baseline is recalculated each run, detection adapts as your traffic patterns change. Sustained shifts gradually become the new normal rather than firing repeatedly.

Types of Anomalies

Visitor Anomalies

Type	Example	Significance
Spike	Far above the usual range	Campaign success? Bot traffic?
Drop	Far below the usual range	Server issue? SEO problem?

Bounce-Rate Anomalies

Type	Example	Significance
Sharp increase	Bounce rate well above baseline	UX issue? Broken page? Bad traffic source?

Value-Score Anomalies

Type	Example	Significance
Sharp drop	Average visitor value falls below baseline	Lower-quality traffic? Targeting change?

For broader observations beyond these three metrics (engagement trends, geographic shifts, opportunities), see the AI Insights you receive alongside anomalies.

Viewing Anomalies

Anomalies and other AI insights live on each website's Insights tab:

Open your website's dashboard from Domains.
Select the Insights tab in the sidebar.
Critical and warning items appear at the top as alerts; the full list sits below.
Click any item to open its details.

Anomaly Details

Each anomaly shows:

Element	Information
Metric	What changed (visitors, bounce rate, or value score)
Actual value	The recorded value for that day
Expected value	The baseline average it was compared against
Deviation	Percentage change from expected
Severity	Info, warning, or critical
Detected	When it was flagged

Sample Anomaly

CRITICAL ANOMALY

Metric: Visitors
Detected: Today
Expected: ~1,100 visitors/day
Actual: 580 visitors
Deviation: -47% from expected

Suggested next steps:
- Check uptime and recent deployments
- Review traffic sources for a drop-off
- Compare against the visitors chart

Acting on Anomalies

When something is flagged, work through it the same way you would any incident.

Step 1: Verify

Confirm the anomaly is real:

Check the underlying metric in the dashboard.
Look at related metrics for the same day.
Rule out tracking or data-collection issues.

Step 2: Understand Scope

Determine what is affected:

Specific pages?
Certain traffic sources?
Mobile vs desktop?
Geographic regions?

Step 3: Find the Cause

Look for correlations:

Recent site or campaign changes?
External events?
Server or uptime issues?

Step 4: Resolve

Once you have looked into it, acknowledge the insight so it moves out of your active list. Acknowledged items stay in your history for reference. See Acknowledging insights for details.

Common Anomaly Causes

Visitor Drops

Cause	Investigation
Server issues	Check uptime monitoring
SEO changes	Search rankings
Ad stops	Campaign status
External events	News, seasonality

Visitor Spikes

Cause	Investigation
Viral content	Social mentions
Bot traffic	Traffic quality, referrers
Campaign launch	UTM parameters
Press coverage	Referrers

Bounce-Rate or Value-Score Changes

Cause	Investigation
UX or page bug	Session recordings, heatmaps
New traffic source	Referrers and UTM parameters
Landing-page change	Recent updates

False Positives

Why They Occur

A day can be flagged even when nothing is wrong:

One-off events (a single big referral day).
A young site with a short baseline.
Genuine but harmless variation.

Reducing Noise

Give the baseline time to fill in (it improves with more daily history).
Acknowledge items that turn out to be expected so they leave your active list.
Treat info and warning items as context rather than emergencies; reserve urgency for critical.

Getting Notified

Anomaly alerts can also reach you by email. Open Settings → Account → Preferences and use the Email notifications section to toggle:

Anomaly Alerts - email about unusual patterns in your analytics
Traffic Spike Alerts - email when your sites see unusual traffic spikes
Weekly Reports - a weekly analytics summary every Monday

These are per-user email preferences. Inside the app, anomalies always appear on the Insights tab regardless of your email settings.

Best Practices

Triage by Severity

Severity	Suggested response
Critical	Investigate promptly
Warning	Review the same day
Info	Note it during your regular review

Keep Notes

When you investigate an anomaly, jot down what happened, the likely cause, and what you did. Over time this builds a useful record of how your site behaves.

Coordinate With Your Team

Share critical anomalies.
Decide who investigates what.
Acknowledge items once they are handled so the active list stays meaningful.