Anomalies detection

You can have Smartlook automatically monitor and alert you when anomalies occur in your events or funnels. You can enable Monitor anomalies in any event or funnel. However, the monitoring is only valid for that funnel or event. If you want to monitor for anomalies in more funnels or events, you have to enable it in each event or funnel.

Enabling anomalies monitoring

To enable anomalies monitoring:

  1. Select the funnel or event you want to monitor.
  2. Click the Anomalies monitoring icon.
  3. Enable Monitor anomalies.
  4. (Optional) Enable Notify me via webhook if you have webhooks set up.
  5. Set the Alert type.
  6. Set what you want to Detect.
  7. Click Save settings.

Event alert types

Alert typeDescription
Automatic detectionAutomatic detection anomalies are determined by comparing changes in the ratio of total event occurences to daily visitors from the previous day with the average of the previous seven days. If the difference fits into the .95 confidence percentile (approximately: mean +/- 2* standard deviations) then a warning email is sent to your address.
Anomaly thresholdYou can set up Anomaly thresholds to be alerted if anomalies fall outside of the predefined threshold value.
Relative percentage changeRelative percentage changes are detected if the conversion rate change is outside of the predefined relative percentage threshold.

Event anomalies

Event anomalyDescription
Surges and dropsDetects both surges and drops in how often the event occurs
SurgesDetects surges in how often the event occurs
DropsDetects drops in how often the event occurs

Funnels alert types

Alert typeDescription
Anomaly thresholdYou can set up Anomaly thresholds to be alerted if anomalies fall outside of the predefined threshold value.
Relative percentage changeRelative percentage changes are detected if the conversion rate change is outside of the predefined relative percentage threshold.

Anomaly notification

If Smartlook detects an anomaly, you receive an automatic email with details related to the anomaly.