When history fails you, borrow from geography

Short summary

When unprecedented shocks break historical data patterns, Airbnb discovered that geography can substitute for time. By tracking how booking lead times recovered sequentially across regions during COVID—North America bouncing back in Dec 2020, Europe following months later—they used earlier-recovering markets as predictors for delayed ones. This 'geographic time machine' approach forecasted demand when local data was unavailable, treating regional lag patterns as transferable signals.

• •Historical forecasting models fail during unprecedented disruptions because they assume future resembles past
• •Geographic lag analysis: earlier-recovering markets predict patterns for markets recovering later
• •Booking lead time compression/recovery timing varies by region, creating transferable predictive signals

Generated with AI, which can make mistakes.