Score each user's chance to subscribe and use local weather to time offers

Data science builds propensity models that merge in-product behavior and known user data into a likely/unlikely-to-subscribe signal. Combined with weather-based targeting (weather as a privacy-forward contextual signal that predicts user action) to decide who sees which subscription packaging and which features are elevated. Moving from mostly anonymous traffic (much from Google search) toward getting users to share an email so targeting can improve. Models are re-evaluated and retuned regularly because economics and competition shift.