Abstract: In certain embodiments, a computer-implemented method includes monitoring, over time, values for a service metric associated providing a computerized service over a communication network and evaluating, according to a statistical model, the values for the service metric to determine whether the values are anomalous values. The statistical model includes a predicted distribution of the values for the service metric and a normal value range within the predicted distribution of the values for the service metric. Anomalous values may be values for the service metric outside the normal value range. The method includes detecting a performance issue with the computerized service and determining, in response to detecting the performance issue, whether one or more of the values for the service metric are anomalous. The method includes automatically setting, in accordance with whether one or more of the values are anomalous, a value of a service level threshold for the service metric.