ic_005 · Industry: Marketing — lead → SQL conversion
Status: ✓ Passing · Last run: 2026-06-07T06:45:28 · Pipeline: v2
· History: 1/1 runs passed (100%)
What was tested
Marketing — lead → SQL conversion Lead-conversion model: leads + touchpoints + opportunities + labels. Signal: high engagement touchpoints drive conversion.
What we planted in the data
- 300 entities
- Signal strength: 0.70 (sigmoid slope multiplier)
What the system did
- Training rows built: 1,200
- Features used in model: 25
- Model selected:
RandomForestClassifier - CV AUC: 0.781
Top features by importance:
touchpoints_demo_request_count_180d(0.160)lead_age_days(0.143)touchpoints_webinar_count_90d(0.134)touchpoints_total_count_365d(0.119)touchpoints_content_download_count_30d(0.091)
What we checked — all assertions passed
| Status | Assertion | Detail |
|---|---|---|
| ✓ | no_pipeline_errors |
clean run |
| ✓ | feature_survival_rate_above |
25/25 features survived (100% ≥ 40%) |
| ✓ | source_tables_contributing_above |
6 distinct source-table prefixes in surviving features (['company', 'lead', 'opp', 'source', 'stage', 'touchpoints']) |
| ✓ | auc_in_band |
AUC = 0.781 inside band [0.55, 0.95] |
How this could have gone wrong
(no assertion descriptions provided)
Why this case matters
Industry-canonical ML problem; multi-table shape mirrors real customer datasets.
Reproducing
# from auto_insight_api/
python -m validation.v2 run ic_005 --pipeline v2 -v
- Case config:
validation/v2/cases/ic_005_marketing_lead_conversion.yaml - Data shape:
industry_marketing_lead_conversion - Analytics type:
ml_binary