Sign in →

Usage Intelligence

Advanced usage analytics — Z-score anomaly detection, linear-regression trend analysis with 30-day projection, and multi-dimension cohort comparison.

Updated 2026-06-15Suggest edits

Usage Intelligence

Usage Intelligence (Intelligence → Usage Intelligence) is the advanced analytics surface for spotting anomalies, projecting trends, and comparing cohorts across your metered usage.

Tabs

Anomaly Detection

Z-score-based spike/drop detection over a 90-day lookback (default threshold 2.5σ, 7-day rolling window). Each anomaly reports:

FieldNotes
TypeSPIKE, DROP, TREND_BREAK
Z-scoreStandard deviations from expected
Severity0.0–1.0
Actual vs expectedObserved value against the baseline
ExplanationPlain-language reason

Trend Analysis

Historical trend plus a 30-day forward projection per metric, with the growth rate, daily growth rate, R² fit score, and a day-of-week seasonality index.

Cohort Comparison

Usage variance across a chosen dimension, with a spread ratio (top vs bottom cohort multiplier):

DimensionGroups customers by
BY_OFFERINGPlan
BY_SIGNUP_MONTHSignup cohort
BY_SUBSCRIPTION_AGETenure

Each cohort bucket reports customer count, total usage, and average / median / p95 per customer.

Usage Intelligence analyzes metered usage patterns. For the storefront visitor funnel, see Storefront Analytics; for cost/margin analytics, see COGS & Margins.