Product update - January 7th, 2026
This release introduces smarter handling of missing data, a clearer way to categorize anomalies, and a new metric for quantifying deviation severity.
Smarter, Noise-Reduced Missing Data Alerting
Our goal is to ensure you are only alerted to issues that truly matter. We've enhanced our missing data policy to drastically reduce false positives caused by expected data pauses.
Old Behavior | New Behavior | Customer Benefit |
Missing data alerts for high-frequency metrics were raised unconditionally after a fixed time (e.g., 30 minutes). | Missing data is now only alerted if the duration exceeds its recently observed historical pattern (within the last 7 days). | Reduced Alert Fatigue: Metrics that predictably stop sending data (e.g., overnight or during scheduled maintenance) will no longer trigger unnecessary alerts. |
Low-frequency metric missing data was limited to only a few customers. | This enhanced, pattern-based missing data detection is now enabled for all high and low-frequency metrics and environments. | Consistent Monitoring: You receive intelligent, pattern-aware missing data alerts across your entire environment. |
*Note: A future release will introduce time-of-day differential policies for even more precise handling of daytime vs. nighttime data patterns.
Clearer Anomaly Classification Flag
To streamline your root cause analysis and filtering, we're introducing a new, dedicated flag for anomaly classification.
Functionality: A new field will be available across the platform to clearly categorize the source of an anomaly.
Categories: This flag will indicate one of two things:
The anomaly is due to Missing Data.
The anomaly is due to a Deviation from Known Behavior (i.e., the value received is outside the normal range).
Customer Benefit: Faster Triage: You can immediately tell whether an alert is a connectivity/data issue or a true performance/behavior issue, allowing you to route and filter alerts with greater precision.
Quantified Deviation Score for Severity
We are introducing a normalized and quantifiable measure of how much an anomalous data point deviates from normal behavior. This metric will significantly enhance your ability to prioritize incidents.
New Field: A new, signed numeric field will be added to our APIs and webhooks.
Calculation: This value is calculated based on the distance of the data point from the closest established baseline, normalized by the width of that baseline.
Example Calculation: If the closest baseline ranges from 5 to 15 (a width of 10), and the received data point is 35:
Deviation Score = (Data Point - Baseline Upper Bound) / Baseline Width
Deviation Score = (35 - 15) / 10 = 2
Value = 0: The data point is within the expected baseline range (not an anomaly).
Positive Value (>0): The value exceeds the upper bounds of the baseline. A higher number indicates a more severe upward deviation.
Negative Value (<0): The value is lower than the lower bounds of the baseline. A more negative number indicates a more severe downward deviation.
Customer Benefit: Objective Prioritization: You can now instantly quantify the severity of an anomaly without needing to understand the underlying metric's scale, allowing for consistent filtering and prioritizing of your most critical deviations.