Meta rebuilt the machine that decides which ad wins each impression, and the results it reports are hard to ignore. Meta says improvements to its ads ranking system drove 8.3% more clicks and a 15.7% lift in conversions on Facebook in the second quarter of 2026. Three upgrades do the work: Andromeda, GEM, and GenRecs. None of them needs a setting switched on. The question is what advertisers should do now that the system reads, predicts, and judges ads differently.
What did Meta actually change?
Meta's system picks an ad in stages. First Andromeda, the retrieval system, narrows tens of millions of eligible ads to a shortlist for each person. Then ranking models pick the winner. GEM is the foundation model that teaches those ranking models. GenRecs is the newest layer: it reads each ad and describes it so the system can understand it. Each upgrade fixes a different job. Andromeda considers more ads, GEM predicts better everywhere, and GenRecs understands new ads faster. Meta announced all three on October 28, 2026.
How does Andromeda consider more ads?
Andromeda builds the shortlist, and its next generation routes each ad to specialized expert models, bringing near ranking-level judgment into retrieval. It scores every eligible ad and surfaces the top candidates at over one billion ads per second. Meta says that since 2025, Andromeda has delivered a 2.4% increase in conversions on Facebook. In the third quarter, Meta made it ten times more complex without increasing computing costs, and Meta reports conversions rose 1.37% on Facebook and 0.24% on Instagram Reels.
What does GEM do now?
GEM, the Generative Ads Recommendation Model, is now the foundation model behind ads recommendations across Facebook, Instagram, Reels, and off-Meta conversions, for every major advertiser goal. It is too large to run on every ad request, so it teaches the real-time ranking models. Meta says that since 2025, GEM has delivered up to an 18.9% conversion improvement and a 12% ad click-through rate improvement. We covered GEM when it arrived in What Is Meta's GEM?. What is new is that it now sits underneath everything, teaching the models that make the final call.
How does GenRecs help new ads?
GenRecs is the newest of the three, and it solves the cold-start problem. It reads each ad's text and images and describes the ad in semantic tokens, brand, color, fit, style, so a new ad can be matched to what the system knows about similar ads. It now enriches Andromeda through generative retrieval. Meta says GenRecs expanded to generative retrieval on Facebook Feed in the third quarter of 2026, increasing conversions by nearly 2% across targeted segments, and that paired with GEM and the Adaptive Ranking Model, it delivered a 0.35% lift across surfaces. The cold start gets shorter, not solved. Meta's ads infrastructure VP told AdExchanger, "You still have to learn."
What should advertisers do about it?
Nothing to switch on. All three reach accounts automatically. Meta says how much an advertiser gets depends on what it gives the system: the outcomes it optimizes for, the quality of the data it shares, and how much optimization it hands over. That points to three practical moves. First, feed the system creative variety, because GenRecs reads the creative itself. Use the Meta Creative Count Calculator to size how many creatives to run. Second, set clear conversion outcomes and share clean conversion data. Third, stop hand-tuning audiences, because Meta's ads infrastructure VP told AdExchanger the importance of targeting is diminishing as ranking improves. Use the Meta Ads Budget Estimator to size the budget for the new ranking.
The honest caveat is that these are Meta's own platform-level figures, not independently verified, and they compare Meta's new models to Meta's old models, not ads to no ads. They are averages, not a promise for any single account.
How we can help
We as your partner manage Meta campaigns as a managed service. We set the conversion outcomes and the data the models learn from, build creative variety so the ranking system has more to work with, plan budgets across Meta surfaces, and measure what the models actually delivered, so the plan is judged on leads, calls, and sales, not on platform-reported lifts.
Takeaway
Meta's ranking system now considers more ads, predicts better everywhere, and understands new ads faster. The advertisers who gain most will be the ones who feed it the best inputs: clear outcomes, clean data, and varied creative. The machine does the selecting. The strategy is still yours.
Are you feeding Meta's ranking system the inputs it rewards, or still tuning the levers it already handles?
Data over opinions.
Frequently asked questions
What are Meta's three new AI innovations in ads ranking?
Andromeda, the retrieval system that builds the shortlist; GEM, the foundation model that teaches the ranking models; and GenRecs, which reads each ad's text and images and describes it in shared tokens so new ads can be matched to what the system knows.
Do I need to change any settings to get these upgrades?
No. Meta says all three reach accounts automatically. What changes is what you feed the system: the outcomes you optimize for, the quality of the data you share, and how much optimization you let the models handle.
Will these changes replace media buyers?
No. The models handle ad selection and audience matching, but they cannot set strategy, decide what a conversion is worth, or build creative. Meta's own ads infrastructure VP told AdExchanger the models cannot set strategy, and that advertisers should spend less time trying to outperform the machine at mechanical levers.
How big are the gains Meta reports?
Meta says its ranking improvements drove 8.3% more clicks and a 15.7% lift in conversions on Facebook in the second quarter of 2026. Those are Meta's own platform-level figures, not independently verified, and they are averages, not a promise for any single account.
Sources
- Meta for Business, Inside the Three New AI innovations in Ads Ranking Improving your Ad Performance
- AdExchanger, Meta Has A New Ad Ranking System Built On LLMs, Plus Upgrades To Andromeda And GEM
Relevant Tools
Want Meta campaigns that feed the ranking system what it rewards?
Get one data-backed marketing insight every week. The same no-fluff analysis behind this article, applied to what is actually moving budgets.
Free forever. Unsubscribe anytime.
Connect with JC