Optimize
Predicted Lifetime Value models, accuracy tracking, and feature importance analysis
Median pLTV
$42.80
+6.2%
vs last periodModel Accuracy
89%
+3.0%
vs last periodMAE
$4.20
-18.4%
improvingR² Score
0.84
+5.0%
vs last periodLTV Curve (90-day, cents)
LTV by Channel (cents)
Model Management
gradient_boost_v3activeGradient Boosting
Trained: 2026-03-01284.0K samples42 features
Accuracy
89%
MAE
$4.20
R²
0.84
Latency
12ms
neural_v2shadowNeural Network
Trained: 2026-03-03284.0K samples42 features
Accuracy
91%
MAE
$3.80
R²
0.87
Latency
28ms
gradient_boost_v2retiredGradient Boosting
Trained: 2026-02-01242.0K samples38 features
Accuracy
86%
MAE
$5.10
R²
0.80
Latency
11ms
Feature Importance (Top 10)
d7_revenue
28%
session_count_d3
18%
subscription_tier
14%
onboard_complete
11%
push_opt_in
8%
country_code
6%
device_model
5%
source_channel
4%
content_interactions_d7
3%
referral_source
3%
Prediction Accuracy by Cohort
| Cohort | Predicted | Actual | Error | Status |
|---|---|---|---|---|
| 2026-W09 | $42.80 | $44.10 | +3.0% | accurate |
| 2026-W08 | $41.20 | $40.80 | -1.0% | accurate |
| 2026-W07 | $39.50 | $38.90 | -1.5% | accurate |
| 2026-W06 | $38.80 | $36.20 | -6.7% | drift |
| 2026-W05 | $37.40 | $37.80 | +1.1% | accurate |
LTV by User Segment
| Segment | Users | % of Total | Avg LTV | Median LTV | Total Revenue | Key Features |
|---|---|---|---|---|---|---|
| Whales | 420 | 2.1% | $284.50 | $242.00 | $119,490 | High D1 spend, multi-purchase, premium tier |
| Dolphins | 3.2K | 16% | $68.20 | $52.40 | $218,240 | Subscription converts, moderate engagement |
| Minnows | 8.4K | 42% | $12.40 | $8.80 | $104,160 | Trial users, occasional IAP, ad-supported |
| Free | 8.0K | 39.9% | $1.80 | $0.40 | $14,364 | No purchase, ad revenue only, low retention |