Use Chat Analytics to detect participation shifts, capacity risks, emerging spam, and moderation workload so you can sustain healthy real‑time conversations.
Message Volume
Total & trend
Participation
Unique senders & activation
Channel Health
Distribution & concentration
Engagement Quality
Reactions & replies depth
Moderation Load
Flagged rate & review pressure
Temporal Patterns
Daily & hourly peaks
Fast Questions You Can Answer
- Is growth broad?
- Quality dip?
- Channel sprawl?
- Safety load?
- Capacity risk?
- Activation health?
Are more unique users sending messages vs just power users ramping?
Daily Operational Workflow
1
Apply Filters
Pick date range (e.g., last 7 / 30 days), segment (environment, region), and exclude internal test channels.
2
Scan KPI Cards
Messages, Flagged Messages, Reactions, Unique Senders—compare % delta vs prior period.
3
Review Volume Trend
Open New Messages by Day to detect bursts or weekend/weekday abnormalities.
4
Check Participation Breadth
Contrast Users by Day with Messages—if messages up but users flat, investigate power user over-reliance.
5
Channel Distribution
Check Channels Count by Type and Messages by Channel for concentration or underutilized types.
6
Moderation Pressure
Flagged Messages & Flagged Rate vs moderation SLA; queue backlog may require staffing shift.
7
Temporal Hotspots
Use Heatmap to validate moderator coverage during peak hours.
8
Drill Power / Risk Users
Use Users table to inspect top senders & high flagged contributors.
9
Log Actions
Document interventions (rate limits, highlights, education) and set follow-up date.
Modules & Interpretation
KPI Cards
High-level snapshot (Messages, Flagged Messages, Reactions, Unique Senders). Deltas contextualize growth vs prior period.New Messages by Day
Volume rhythm; use for feature launch impact and anomaly detection. Gentle cyclic pattern is normal; abrupt plateau indicates engagement stall.Channels Count by Type
Composition (public, private, broadcast). Skew heavily toward one type may limit discovery or create moderation blind spots.Users by Day
Active unique senders; rising volume without matching unique sender growth implies intensity, not breadth.Total Messages by Type / Channel Distribution
Identifies reliance on a small set of channels. Over-concentration risks single-point community health issues.Top Group Chat Channels Leaderboard
Sort by Members, Messages, Engagement Rate (messages per member), or Flagged Rate to surface exemplars or risk clusters.Messages Heatmap
Hourly/daily density. Align moderator shifts and system scaling policies with dark (peak) cells.Key Metrics & Formulas
| Metric | Definition | Formula (Illustrative) | Why It Matters | Action Trigger |
|---|---|---|---|---|
| Messages | Total messages sent | count(messages) | Overall volume | Drop >15% w/w |
| Unique Senders | Distinct users sending ≥1 message | distinct(user_id) | Participation breadth | Flat while messages rise |
| Messages per Active User (MPAU) | Avg intensity per sender | Messages / Unique Senders | Detect over-reliance | Sharp rise + flat senders |
| Reactions | Total reactions on messages | count(reactions) | Lightweight engagement | Reactions/Message down |
| Reaction Rate | Reactions per message | Reactions / Messages | Content resonance | Below baseline band |
| Flagged Messages | Messages flagged by systems/users | count(flag events) | Moderation workload | Spike >25% w/w |
| Flagged Rate | Flagged / Messages | Flagged / Messages | Safety cleanliness | > threshold (e.g., 1–2%) |
| New Channels | Channels created period | count(channel_create) | Organic growth / fragmentation | Spike + falling engagement rate |
| Channel Concentration Index | Engagement distribution (simplified HHI) | Σ (channel_share^2) | Diversity of conversation | > prior period + rising churn |
| Messages per Channel | Avg messages per active channel | Messages / active_channels | Capacity & sprawl measure | Decline + rising channel count |
| Peak Hour Factor | Peak hour messages vs avg hour | peak_hour / avg_hour | Staffing & infra planning | Factor rising > target |
| Newly Activated Chat Users | First-time senders in period | distinct(new_sender_ids) | Feature adoption | Activation drop consecutive weeks |
| Retained New Senders (7d) | New senders returning within 7d | returning_new / new_sender_ids | Early retention | < benchmark band |
Participation & Quality Lenses
- Acquisition
- Activation
- Retention
- Quality
- Safety
- Load & Capacity
Growth in newly activated chat users.
Benchmark & Governance Strategy
Baseline Collection
Baseline Collection
Capture 30 rolling days after stable launch; derive median & IQR for each core metric.
Adaptive Thresholds
Adaptive Thresholds
Alert if metric leaves median ±1.5 * IQR (robust vs outliers).
Cohort Drift
Cohort Drift
Track new vs established communities; avoid mixing maturity profiles.
Quality Gate
Quality Gate
Require Reaction Rate & Retained New Senders above thresholds before promoting channels.
Moderation SLA
Moderation SLA
Define max backlog aging (e.g., 2h). Breach triggers escalation & auto-priority ordering.
Early Warning Signals
- Power User Dependence
- Fragmentation
- Spam / Flood
- Moderation Overload
- Quality Decline
- Adoption Stall
MPAU rises, Unique Senders flat.
Troubleshooting
| Symptom | Likely Cause | Investigation | Resolution |
|---|---|---|---|
| Volume up; engagement flat | Few users posting heavily | Distribution of messages per user | Encourage broader participation; prompts, onboarding nudge |
| Flagged rate surge | Spam / abusive burst | Review top flagged channels & user IDs | Apply rate limits; tighten filters; temp-ban offenders |
| Reaction rate drop | Content low relevance / UI friction | Compare reaction latency post message | Improve quick reaction UI; highlight engaging threads |
| New channels spike; avg messages/channel down | Fragmentation / channel creation sprawl | Correlate channel age vs activity | Introduce channel creation guidelines & approval |
| Moderation backlog aging | Understaffed peak or automation gap | Backlog aging metric by hour | Add on-call; refine auto-flag thresholds |
| Low new sender retention | Poor onboarding into existing channels | Funnel: Signup → First message → Return visit | Provide channel suggestions & starter messages |
| Peak hour factor rising | Emerging synchronized usage cluster | Compare heatmap vs previous weeks | Pre-scale infra; adjust sharding or throughput configs |
Related
Activity Analytics
User activity baseline
Social Analytics
Community performance
AI Social Insights
Qualitative topics & sentiment
Raw Data Export
Custom message logs
Need custom spam detection dashboards or extended retention windows? Contact support for advanced analytics enablement.