Docs

Anomaly Detection

The Anomaly Detection agent watches your metrics around the clock and alerts you when something unusual happens — a revenue spike, a download drop, a conversion rate shift, or any other significant change from normal patterns.


How it works

The agent runs hourly by default, analyzing your historical data to establish what's "normal" for each metric. It accounts for trends, seasonality, and day-of-week patterns. When a new data point falls significantly outside the expected range, it's flagged as an anomaly.

The system compares each value against multiple time horizons:

  • Same day last week
  • Same day last month
  • 7-day rolling average
  • 30-day rolling average

This multi-scale approach catches both sudden shocks and gradual shifts.

Minimum requirement: 30 days of historical data. New apps or newly connected data sources may not trigger anomalies until enough history accumulates.


Anomaly types

TypeDescription
SurgeMetric significantly higher than expected (e.g., revenue spike)
DeclineMetric significantly lower than expected (e.g., download drop)

Severity levels

SeverityWhat it means
LowMild deviation — worth watching
MediumNotable deviation — review soon
HighSignificant deviation — investigate today
CriticalExtreme deviation — immediate attention recommended

What gets monitored

By default, the agent monitors these core metrics:

  • Proceeds — net revenue after Apple/Google fees
  • Total revenue — gross revenue across all sources
  • Total downloads — app installs
  • Active subscriptions — current paying subscribers
  • New trials — subscription trial starts

You can configure additional custom metrics (e.g., crashes, sessions, ad spend) in Settings → Monitoring.


Alerts

When an anomaly is detected, you're notified through your preferred channels:

  • Email — real-time or daily digest
  • Slack — posted to your configured channel
  • Discord — posted to your configured channel

Configure which severity levels and metric types trigger notifications in Settings → Notifications.


Viewing anomalies

Go to Monitoring in the sidebar to see all detected anomalies. Each anomaly shows:

  • The metric and date
  • Actual value vs. expected value
  • Deviation percentage
  • AI-generated explanation of what likely happened
  • Suggested next steps

You can filter by severity, type (surge/decline), app, metric, and date range.


Dismissing anomalies

Once you've reviewed an anomaly and understand the cause, dismiss it to keep your dashboard clean. Dismissed anomalies are hidden from the active view and moved to the archive.


What's free vs. paid

FeatureFreePro ($50/month)
View all monitoring metrics and dashboardsYesYes
View anomaly history and alert logYesYes
Anomaly notifications (email, Slack, Discord)YesYes
AI-generated explanations of anomaliesYes
AI-suggested next stepsYes

See Pricing for the full breakdown.

PreviousAds Agent
NextProduct Testing