Optimixed Search News

Meta reports GEM unified ads model boosted conversions 18.9% across its apps

45
News

TLDR

Meta says its new GEM (Generalized Embedding Model) unified ads prediction system lifted conversions 18.9% and click-through rates 12% across Facebook, Instagram, and other apps.

Optimixed Overview

Optimixed's original report on this story.

Meta announced that its GEM (Generalized Embedding Model), a unified machine-learning system trained on thousands of GPUs to predict ad performance across all Meta surfaces, achieved a 18.9% conversion lift and 12% click-through rate increase in testing. The model consolidates ad prediction into a single system rather than maintaining separate predictions for each platform, though the article does not detail what transparency or performance-verification tools advertisers have access to.

Was this useful?
Share

Related articles

35
Product announcement

Google has launched Universal Cart, a shopping feature now available in U.S. Search and the Gemini app.

Paid Search & AdvertisingEcommerce MarketingAI & Machine Learning in SEO
35
News

Google introduced Vehicle Top Ads, a new search ad format in beta that lets automotive manufacturers display vehicles with rich visuals and details directly in search results.

Paid Search & AdvertisingSEO News & Algorithm Updates
35
News

Addressability and targeting restrictions re-emerged as a major concern at Prebid Summit, driven by Apple's iOS 27 privacy changes.

Paid Search & AdvertisingSEO News & Algorithm UpdatesTools & Platforms
35
Product announcement

Microsoft Advertising published seasonal guidance on reaching holiday shoppers via AI-optimized search, shopping, and audience ads.

Paid Search & AdvertisingAI & Machine Learning in SEOEcommerce Marketing