Google will
now have the ability to predict consumer behavior changes hours or days before they happen to more accurately target ads.
On Thursday, Google DeepMind introduced WeatherNext 3, a weather forecasting model that generates predictions every hour by
drawing on live satellite data rather than waiting for government-produced datasets that update every six hours.
advertisement
advertisement
Previously, weather targeting had operated via broad city or postal codes. Data
from WeatherNext 3 will allows Google to target ads based on micro-climates such as a sudden rainstorm in one specific neighborhood while a neighboring suburb remains sunny.
The blog post did
not explain whether Google would use this in advertising and if so, how it would be used.
Google did explain that WeatherNext 3 will now power forecasts in Google Search, Gemini App, Google Maps and Google Maps Platform.
It is now integrated across Search, Gemini, Maps, Google Maps
Platform, and Cloud.
The forecast weather model includes real-time satellite data, with higher resolution, precise precipitation forecasting, and clean energy variables.
Developers and researchers also can access real-time data via BigQuery, Earth Engine, and Google Cloud Storage
(GCS).
Weather influences billions of decisions daily, according to Google's blog post.
While some are simple, others are far more consequential. Wind, rain, and
extreme weather events such as heatwaves and droughts have cascading impacts across agriculture, global supply chains, clean energy production and national economies.
Weather also has an impact on what consumers purchase, although Google did not mention how the data from
the model will be integrated into advertising campaigns to more accurately serve ads based on predictions what consumers need.
Creative adjustments based on weather could be done
in real time. If the 5km grid detects a sudden increase in temperature, a fast-food brand's ad could automatically switch from promoting hot coffee to iced coffee for users in that geographic
area.
If high wind speeds combined with sudden rain are detected, this could instantly trigger ads for local hardware stores selling emergency tarps, or ride-share apps offering rides during a
downpour.
Some businesses see massive revenue shifts based on minor weather changes. For example, humidity data could trigger ads for anti-frizz hair products, rain jackets, or athletic
gear.
Weather changes so rapidly in Wyoming that the joke has always been that if you do not like the weather, wait a minute and it will change.
Google
DeepMind's weather model learns directly from real-time observations, enabling it to provide timely and more localized predictions for the weather events that impact people.
By using raw satellite data to produce a forecast hourly in high resolution, the model makes reliable forecasts accessible across Google products worldwide.
WeatherNext 3 can
visualize key surface variables such ass temperature and moisture at a 5-kilometer resolution, other surface variables at 10 kilometers, and atmospheric variables, like wind speed, at 25
kilometers.
This provides a global weather picture five times sharper than Google DeepMind's previous model, WeatherNext 2, which produced forecasts on a 25-kilometer grid in 6-hour
increments.