To balance query performance with accuracy, the # Sessions metric (representing the distinct count of application Session IDs) for Alert, App, Incident, Page, and Transaction Events utilizes the HyperLogLog (HLL) algorithm.
This probabilistic data structure returns an approximate cardinality of unique session IDs. While this ensures high-speed query execution, it may result in a margin of error of approximately 0.5% compared to a deterministic count. Internal statistics indicate that for 95% of use cases, there is no discernible difference between the HLL approximation and the exact count.
If a strictly accurate count is required for specific auditing purposes, please narrow your scope using precise filters to limit the dataset size before querying.