Mobile F2P Analytics — Homescapes-type
Synthetic dataset · 50,000 players · Portfolio case study
50K
Players
6
Cohorts
2
A/B Groups
D1 Retention
51.4%
Overall avg
D7 Retention
27.8%
Overall avg
D30 Retention
10.2%
Modeled proxy
A/B Uplift
+9.1%
p <0.001
Avg LTV (D30)
$14.2
Blended
Whale Share
5%
62% of revenue
Retention Curves — Control vs Treatment (D0–D30)
Cohort Retention Heatmap (%)
CohortWeek 1Week 2Week 3Week 4
Session Depth Distribution
A/B Test Summary
Control D7 Ret.
25.7%
Treatment D7 Ret.
28.0%
Relative Uplift
+9.1%
p-value
<0.001
95% CI
[+7.4%, +10.8%]
Z-score
8.34
Sample (ctrl)
25,021
Sample (treat)
24,979
Statistical Power vs Sample Size
A/B Retention by Day
D7 Retention by Segment
Behavioral Profile Radar
Segment KPI Breakdown
SegmentPlayersShareD7 RetAvg SpendRevenue Share
LTV Proxy Model — Predicted vs Actual (30-day)

Model Notes

• LTV proxy uses D1/D7 retention + session depth as early signals

• Whale segment predicted within ±8% of actual

• Ghost segment has high uncertainty due to near-zero engagement

• Model RMSE: $4.2 blended across segments

Early signal correlations:

• D1 session count → D30 spend: r = 0.61

• Levels cleared D1-3 → D30 LTV: r = 0.74

Revenue Share by Segment
Level Progression Funnel
Churn Risk Signals
Key insight: Players with <2 sessions in first 3 days have 4.2× higher churn probability. Recommend push notification trigger at 36h inactivity.
A/B Guardrail Metrics