Lifecycle States

Understanding the 9 lifecycle states, how transitions work, and what they mean for your community

👑 Pro

State Transition Flow

Typical Member Journey

NewActivatingActiveCore

Churn Path

ActiveDriftingAt RiskFragileDormantLost

Recovery Path

DormantRevivedActive

Note

Members can skip states in both directions. A highly active new member may jump directly from New to Active, and a member who suddenly stops all activity can move from Core to At Risk in a single compute cycle.

The 7-Layer Decision Engine

Each compute cycle, every member passes through seven evaluation layers in order. The first layer that produces a definitive result determines the state:

L1Forced Overrides

Handles absolute conditions: members inactive beyond the lost threshold are forced to Lost, beyond dormancy threshold to Dormant. These override all other logic.

L2Inactivity Detection

Checks days since last activity against configured thresholds. If inactive beyond the inactivity threshold, assigns At Risk or Fragile based on severity.

L3Re-engagement Signals

Detects when previously inactive members return. If recent activity is found for a member in a declining state, transitions to Revived to give them a fresh evaluation window.

L4Core Qualification

Evaluates whether a member meets Core status: high health score, strong social connections, consistent activity over 30+ days, and top percentile engagement.

L5Engagement Gates

Checks health score against the engagement threshold to determine if a member qualifies as Active. Also evaluates trending direction to catch members who are declining from Active.

L6Activation Progress

For newer members, evaluates onboarding progress. Members showing early engagement signals (first messages, voice, reactions) are assigned Activating.

L7Drift Detection

Catch-all layer for members who don't clearly fit other states. Evaluates health trend direction — declining trends result in Drifting, rising trends in Activating.

Tip

Each state decision includes a human-readable explanation stored as lastExplanation. View it in the dashboard or via /lifecycle member to understand exactly why a member was assigned their current state.

Confidence Bands

Every state assignment includes a confidence score (0.0 to 1.0) and a confidence band that tells you how reliable the assessment is:

Insufficient

Confidence below 25%

Not enough data to make a reliable assessment. Typically seen in the first 24-48 hours.

Low

Confidence 25% - 50%

Some data available but limited. State may change significantly as more activity is observed.

Moderate

Confidence 50% - 75%

Reasonable data coverage. State assignment is fairly reliable with some potential for adjustment.

High

Confidence above 75%

Strong data foundation. The state assignment is very likely accurate and stable.

Warning

When confidence is Insufficient, take state assignments with a grain of salt. MLI needs at least a few days of data before it can make reliable assessments. Avoid making moderation decisions based on low-confidence data.

Score Dimensions

Each member has multiple score dimensions that feed into the overall health score:

Health Score (0-100)

The primary composite score. Combines engagement, social, and consistency metrics into a single number that represents overall member health.

Engagement Score (0-100)

Measures activity volume and consistency: messages, voice time, reactions, and how evenly they're distributed over time.

Social Score (0-100)

Measures social integration: replies received, reactions from others, co-activity with other members, and reciprocity of interactions.

Importance Score (0-100)

Measures the member's impact on the community: influence on others' activity, role level, channel diversity, and overall contribution breadth.

Risk Score (0-100)

Aggregate risk level from all active risk families. Higher risk scores indicate more churn signals detected.

Priority Score (0-100)

Combines importance and risk to identify which at-risk members deserve attention first. High-importance, high-risk members rank highest.

Tip

In the dashboard, you can sort the member table by any score dimension. Sort by Priority Score (descending) to find the members who most need your attention — they're important to the community AND showing risk signals.