Anthony McGrathMeta Ads Ranking Now Powered by Large Language Models
Meta introduces a new Meta ads ranking system using large language models to improve ad targeting and boost conversions.
Quick answer
Meta now uses large language models to rank Meta ads, scoring billions of ads per second for better targeting and increased conversions, easing campaign management.
Key takeaways
- Meta scores billions of ads per second with LLMs for relevance
- New systems shorten learning time for new ads with semantic analysis
- Less manual audience targeting needed as AI selects best ads automatically
Infographic. This infographic shows the key points from meta ads ranking now powered by large language models.
Meta has rolled out a new ad ranking system using large language models, or LLMs, to score over a billion Meta ads every second for relevance. Announced on 28 October 2026 and covered by AdExchanger, this update is part of three upgrades that improve how ads are chosen for you or your customers on Facebook and Instagram.
The system starts with Andromeda, which narrows millions of ads to a manageable shortlist tailored to each user. Then a more powerful ranking model picks the best ad to display. Using LLMs, the new system understands ads by their key features, like brand or style, not just campaign labels that often change. This helps new ads get a quicker start, shortening the learning phase where ads prove their worth.
The addition of GenRecs, a generative recommendation model, predicts relevant ads even with little engagement data. This improves the quality of ads shown and lessens the cold-start challenge. Meta says these upgrades brought an 8.3% rise in clicks and a 15.7% jump in conversions on Facebook during Q2.
What does this mean for your marketing?
Meta’s new ranking system means you don’t have to invest as much time splitting audiences or guessing what might work. The AI handles much of the audience targeting and ad selection automatically. You should focus instead on clear strategy and creative quality.
If you run Meta ads, test different creatives regularly to give the system data to learn. Also, keep your product descriptions clear since LLMs analyse text and images closely. Want to check your Meta ad budgets or calculate expected returns? Try my free Meta ads calculator to plan smarter campaigns this week.
Keeping pace with changes like this can significantly boost your ROI without more manual work.
Questions and answers
What is Meta’s new ad ranking system?
Meta’s new system uses large language models to analyse ad features like brand and style, not just campaign tags. This helps rank ads more precisely and improves performance.
How do these upgrades affect my Meta ads?
They improve ad targeting and conversions automatically, reducing the need for manual audience segmentation and letting your ads reach the right people faster.
What is the cold-start problem in advertising?
It’s when new ads lack engagement data, making it hard to predict their success. Meta’s new semantic model gives new ads a head start by comparing them to similar existing ads.
