Automatically catch spend spikes with Anomaly Detection

Built-in anomaly detection is now available in monitors, automatically accounting for seasonal spending patterns. This new feature ensures you’re alerted only to genuine anomalies, with customizable sensitivity to minimize false positives.

Our existing monitors allow users to set static thresholds to alert them when resource spend exceeds a predefined amount. However, determining the right threshold can be challenging—especially when spending follows seasonal patterns. For example, if a workload is expected to spike every Monday but not on other days, a threshold set to accommodate that pattern might miss anomalies on other days.

To solve this, we’ve integrated built-in anomaly detection into monitors. This new feature uses an algorithm that automatically accounts for seasonality in your spend data, ensuring you’re alerted only to genuine spending anomalies while filtering out regular, predictable fluctuations.

SELECT Automatically catch spend spikes with Anomaly Detection

Additionally, you can fine-tune the algorithm’s sensitivity, allowing you to strike the perfect balance between rapid detection and minimizing false positives according to your own needs.

SELECT Automatically catch spend spikes with Anomaly Detection

Other Things We shipped

🚀 To clarify user management actions, we've introduced a SELECT Organization level access role. This controls destructive actions like removing a user from your SELECT Organization. Any pre-existing Snowflake Organisation roles have been migrated into SELECT Organisation roles to ensure no loss in privileges.