What does a Responsive Display Ad use in its machine-learning model to determine the optimal combination of assets for your ad slot?

Responsive Display Ads leverage machine learning to combine assets (headlines, descriptions, images, logos) into optimal ad variations. This optimization is primarily driven by "Predictions built from your performance history." Google's algorithms analyze how your specific assets have performed in the past across various contexts to predict which combinations are most likely to succeed for a given ad slot. This personalized approach ensures the ads are tailored to your campaign's unique data. While industry benchmarks and aggregated data might inform the initial model, the ongoing optimization heavily relies on your account's historical performance. CRM data is external and not directly used by the Google Ads machine-learning model for asset combination.