Docs/Retention Engine/Send Time Optimization

Send Time Optimization

Automatically deliver messages at the optimal time for each player based on their behavior patterns.

8 min read

Overview

Send Time Optimization (STO) uses machine learning to determine the best time to reach each player. Instead of blasting messages at fixed times, Ilara delivers when each player is most likely to engage.

Results

Studios using STO see 20-40% higher open rates compared to fixed-time sends.

How It Works

Learning Phase

Ilara analyzes historical engagement data to build a model for each player:

Session start times and durations
Push notification open times
Email open and click times
In-app message engagement
Time zone and locale data

Prediction

For each message, Ilara predicts the optimal send window:

Hour of Day: When player typically plays
Day of Week: Active days vs. dormant days
Session Proximity: Just before expected session
Fatigue Avoidance: Not too soon after last message

Enabling Send Time Optimization

Dashboard

Go to Retention → Campaigns → Select Campaign
Under "Delivery", choose "Optimized Send Time"
Set your send window (e.g., 9 AM - 9 PM)
Save and activate

API

Campaign with STO

json

POST /v1/notifications/campaigns
{
  class="code-string">"name": class="code-string">"Daily Reward Reminder",
  class="code-string">"segment_id": class="code-string">"active-players",
  class="code-string">"template": {
    class="code-string">"title": class="code-string">"Your daily reward is ready!",
    class="code-string">"body": class="code-string">"Don't forget to claim your free gems today."
  },
  class="code-string">"schedule": {
    class="code-string">"type": class="code-string">"recurring",
    class="code-string">"frequency": class="code-string">"daily",
    class="code-string">"send_time_optimization": {
      class="code-string">"enabled": true,
      class="code-string">"window_start": class="code-string">"09:00",
      class="code-string">"window_end": class="code-string">"21:00",
      class="code-string">"timezone": class="code-string">"player_local",
      class="code-string">"fallback_time": class="code-string">"12:00"
    }
  }
}

Configuration Options

Option	Description	Default
`enabled`	Enable STO for this campaign	false
`window_start`	Earliest send time (HH:MM)	08:00
`window_end`	Latest send time (HH:MM)	22:00
`timezone`	player_local or specific zone	player_local
`fallback_time`	Time if no data available	12:00
`min_gap_hours`	Min hours between messages	4

Quiet Hours

Always set reasonable send windows. Avoid sending during typical sleep hours (10 PM - 8 AM local time) unless your game has a specific night-owl audience.

Player Preferences

Ilara detects and respects player engagement patterns:

Pattern	Detection	Optimization
Morning Player	70%+ sessions before noon	Prioritize 8 AM - 11 AM
Evening Player	70%+ sessions after 6 PM	Prioritize 6 PM - 9 PM
Commuter	Sessions at 8 AM and 6 PM	Target commute times
Weekend Warrior	80%+ sessions Sat/Sun	Increase weekend sends
Night Owl	Sessions after 10 PM	Extend window, respect sleep

ML Model Details

Features Used

Historical opens: When player opened past notifications
Session patterns: Time-of-day and day-of-week activity
Response velocity: How quickly player acts on messages
Device context: Mobile vs. desktop engagement
Content type: What types of messages get engagement

Cold Start

For new players without history, Ilara uses:

Population-level patterns from similar players
Time zone-based defaults
Initial session timing as baseline
Rapid learning from first few interactions

Analytics & Reporting

Track STO performance in the dashboard:

Key Metrics

Metric	Description
Open Rate Lift	STO vs. fixed-time open rate improvement
Optimal Time Distribution	When messages are actually sent
Engagement by Hour	Open/click rates by time of day
Prediction Accuracy	How often optimal time was correct

STO Analytics Response

json

GET /v1/notifications/campaigns/{id}/sto-analytics
 
{
  class="code-string">"campaign_id": class="code-string">"camp_123",
  class="code-string">"period": class="code-string">"last_7_days",
  class="code-string">"metrics": {
    class="code-string">"total_sent": 45000,
    class="code-string">"open_rate": 0.32,
    class="code-string">"baseline_open_rate": 0.24,
    class="code-string">"lift_percentage": 33.3,
    class="code-string">"peak_send_hours": [9, 12, 18, 20],
    class="code-string">"prediction_accuracy": 0.78
  }
}

Best Practices

Set reasonable windows: 12-14 hour windows work best
Respect quiet hours: Don't wake players up
Allow learning time: Models improve over 2-3 weeks
Segment appropriately: Different segments may need different windows
Monitor fatigue: Watch for declining engagement over time

Next Steps

Multi-Channel Setup — Coordinate across channels
AI Message Generation — Personalize content
Analytics Dashboard — Track performance

AI Message Generation

Multi-Channel Setup