Weather triggers as features
Hail, wind, and storm events drive immediate roofing demand. The platform's home-services model family includes weather-trigger signals as direct features. Behavioral consideration signals (insurance claim research, contractor evaluation behavior) downstream of weather events further sharpen the cohort.
Geo-temporal modeling
Demand operates at the ZIP and DMA level, not the state level. Geo-temporal modeling produces ZIP-level probability cohorts that materially outperform DMA-level targeting on cost-per-appointment.
Decision-window scoring
Post-event decisions cluster in 14–60 day windows depending on insurance dynamics and contractor availability. Window-aware targeting concentrates media in the highest-probability window.
Calibrated decay reference
Signal half-life — production model
Conversion velocity reference
Predictive cohort vs. cold list
Citations
- · NOAA Storm Events Database — public weather-event dataset.
- · Insurance Information Institute — Catastrophe Loss Data, 2024.