anomaly-scan

Marketing performance anomaly detection scans connected platforms to identify statistically significant deviations in traffic, spend, CPA, and conversions by comparing current metrics against rolling historical baselines. It applies sensitivity thresholds and time-window controls to flag issues requiring fast diagnosis.

To diagnose sudden CPA spikes and traffic drops, compare recent metrics against rolling performance baselines using statistical thresholds. This approach correlates deviations with recent executions and known events, categorizing findings as Critical, Warning, or Info with prioritized next actions.

Yes, you can monitor deliverability collapse and budget overruns by comparing current platform metrics against rolling historical baselines. The anomaly scan applies seasonality and day-of-week adjustments with sensitivity controls to identify deviations from established performance baselines.

Finding the root cause of conversion changes involves correlating detected anomalies with recent executions and known events like outages or algorithm changes. The scan outputs categorized findings with probable causes and prioritized next actions for fast diagnosis.

Yes, cross-platform anomaly detection requires loading active brand context and pulling current platform metrics via MCPs. The scan then runs scripted baseline retrieval and anomaly detection with sensitivity and time-window controls to output categorized findings.

Statistical thresholds for anomaly detection rely on rolling historical baselines, so limitations arise when insufficient historical data exists to establish accurate performance baselines. Sensitivity and time-window controls must be properly configured to avoid false positives during expected seasonal fluctuations.