Meta processes over 15 trillion ad ranking predictions a day. Most marketers still brief creative like it’s 2019. If you haven’t heard of Andromeda, Lattice, or GEM, your media buyer probably has, and your creative team is likely working against the machine instead of with it.
Meta’s Advantage+ suite isn’t one algorithm. It’s a stack of three distinct systems, each making different decisions about your ads, and each rewarding a different kind of creative input. Understanding the split is no longer optional for anyone briefing paid social work.
Why This Matters to Brand Marketers Right Now
For years, “the algorithm” was a black box brands treated as one monolithic thing. That mental model is now actively costing budget. Meta has been quietly rebuilding its ad delivery infrastructure since 2023, and by now the architecture has stabilized enough that performance differences between well-briefed and poorly-briefed campaigns are measurable in the double digits.
The three systems worth knowing:
- Andromeda — a retrieval engine that handles ad recall, pulling relevant ads from massive inventory using deep neural retrieval instead of simple rule-based filtering.
- Lattice — the unified ranking architecture that scores and sequences ads across Feed, Reels, and Stories using shared model infrastructure rather than siloed, surface-specific models.
- GEM (Generative Ads Model) — the newest layer, a generative recommendation model that predicts engagement and outcomes using foundation-model techniques borrowed from large language models.
None of this is theoretical plumbing you can ignore because “the platform handles it.” Each layer changes what creative inputs actually get rewarded, and briefs written for the old delivery logic are leaving performance on the table.
Brands still briefing for “the Facebook algorithm” as a single entity are optimizing for a system that no longer exists in that form. Andromeda, Lattice, and GEM each reward different creative signals — and most brief templates were built before any of them existed.
Andromeda: Retrieval Is Where Your Ad Gets a First Chance
Andromeda governs the retrieval stage, the moment Meta decides which ads from a pool of millions even get considered for a given impression. Before Andromeda, retrieval relied heavily on lightweight, rule-based filters that couldn’t process much nuance. Now it’s a deep learning retrieval system capable of understanding semantic relevance between ad content and user context at massive scale.
Practically, that means an ad with rich, specific creative signals (clear product framing, distinct visual identity, well-tagged assets) has a better shot at surviving retrieval than a generic, templated variant. Vague creative doesn’t just underperform, it may not even get considered.
This is where asset diversity earns its keep. Meta has said publicly that advertisers running Advantage+ campaigns see meaningful lift when they supply multiple creative variants rather than one “hero” asset. Andromeda needs options to retrieve from. Starve it, and you’re starving your own reach.
What this means for briefing: stop writing briefs that produce one polished ad. Write briefs that produce a creative system: 5-8 variants covering different hooks, formats, and value propositions, all tagged with clear, structured metadata. If your creative team is still handing off a single 15-second cutdown and three static banners, you’re feeding a retrieval engine that thrives on volume and specificity almost nothing to work with.
Lattice: One Model, Every Surface, Your Cross-Format Creative Needs to Match
Lattice replaced Meta’s older surface-by-surface ranking models with a unified architecture that shares learning across Feed, Reels, Stories, and Marketplace. Before Lattice, a Reels-optimized model and a Feed-optimized model didn’t talk to each other. Now ranking signal from one surface can inform delivery on another.
The upside for brands: less fragmentation, more consistent optimization. The catch: creative that’s built for a single surface and force-fit into others gets penalized by a ranking system that’s now comparing it against natively-built cross-format content in the same pool.
A static image resized into a 9:16 Reels placement with black bars isn’t just an aesthetic problem anymore. It’s a ranking signal problem. Lattice is trained on engagement patterns across formats, and it can tell the difference between content built for vertical, sound-on, fast-scroll consumption versus content that was clearly designed for a different context and dumped into the placement.
This has direct implications for how creative and media teams should collaborate. If your creator briefs still separate “Feed assets” and “Reels assets” into different production tracks with different budgets and different timelines, you’re building for a ranking model that no longer treats those surfaces as separate contests.
GEM: Where Generative Prediction Changes the Performance Math
GEM is the most consequential of the three for brands, because it changes what “creative testing” even means. Instead of relying purely on historical engagement data to predict how an ad will perform, GEM uses generative modeling techniques, similar in spirit to how large language models predict the next token, to predict engagement and conversion likelihood for combinations Meta hasn’t seen much of before.
Meta has reported that GEM contributed to double-digit improvements in ad recommendation quality in its early rollout, according to details shared via Meta for Business. That’s a meaningful jump for a platform already handling trillions of daily predictions.
For brand marketers, the implication is subtle but important: GEM is better at recognizing novel creative patterns than legacy models were. It’s less reliant on “this worked before, so it’ll work again” and more capable of extrapolating from creative structure itself, pacing, hook placement, visual composition, even if that exact combination hasn’t run before.
That’s genuinely good news for brands willing to test unconventional creative. It’s bad news for brands still running the same three ad templates from last year and wondering why performance has plateaued. GEM rewards creative novelty within a coherent brand framework. It does not reward creative laziness dressed up as “brand consistency.”
GEM’s generative approach means Meta’s system can now recognize why a creative pattern might work, not just that it has worked before. Brands that keep recycling last year’s top performer are optimizing for a prediction model that’s already moved past that logic.
How to Actually Rewrite Your Creative Brief
None of this requires a total teardown of your creative process. It requires updating the brief template to match how the delivery system actually works in 2026. Here’s the practical shift:
- Volume with intent, not volume for its own sake. Brief for 6-10 distinct creative concepts, not 10 minor variations of one concept. Andromeda needs semantic diversity, not color-swap duplicates.
- Native-first for every surface. Require separate shoots or edits for vertical, sound-on formats rather than repurposed horizontal assets. Lattice’s cross-surface ranking punishes the shortcut.
- Build in structured metadata. Clear product tagging, accurate captions, and consistent naming conventions help retrieval systems understand what an asset actually is. This is not busywork, it’s a ranking input.
- Leave room for the unfamiliar. Set aside at least one test slot per flight for creative that doesn’t resemble your historical top performers. GEM is built to find signal in patterns you haven’t tried yet.
- Shorten your review cycle. If your internal approval process takes three weeks, you’re not going to keep pace with a system that rewards frequent creative refresh. This is as much an operational problem as a creative one, and it’s worth auditing alongside broader governance and approval workflows.
Agencies briefing creator content for Advantage+ placements should apply the same logic. A creator deliverable brief that specifies platform-native vertical formats, varied hook styles, and structured captions will retrieve and rank better than a brief that just asks for “a few UGC-style videos.” The tagging and metadata discipline that makes AI systems efficient elsewhere in your stack applies directly here too.
It’s also worth remembering that automation in ad delivery still needs human checkpoints. Meta’s systems are optimizing for engagement and conversion signals, not brand safety or message accuracy. Teams that have audited AI-generated creative approval gaps elsewhere in their stack should apply the same scrutiny to Advantage+ output, particularly as generative elements creep further into ad creation tools.
What About Measurement?
One underrated consequence of the Andromeda-Lattice-GEM stack: attribution gets murkier, not clearer. Because ranking and retrieval now share signal across surfaces and predict engagement generatively, isolating “why did this ad perform” becomes harder using last-click logic alone.
Brands relying on legacy attribution models to explain Advantage+ performance are often working from incomplete data. This is the same identity and measurement problem showing up across the ad ecosystem, and it’s worth pairing your creative brief overhaul with a broader look at identity resolution gaps that distort how you read platform-reported results. According to eMarketer, marketers citing measurement confidence as a top challenge in automated ad environments has climbed steadily, and Meta’s architecture shift is a direct contributor.
Practically: lean on Meta’s own reported lift metrics with some skepticism, cross-reference with your own conversion data, and don’t assume platform-attributed performance maps cleanly to incremental revenue. That’s not a new problem, but the generative layer makes it slightly worse before tooling catches up.
The Takeaway
Audit your next Advantage+ brief against three questions: does it supply enough creative volume and diversity for Andromeda to retrieve from, does it treat every surface as native rather than repurposed for Lattice, and does it leave room for the unfamiliar so GEM has something new to learn from? If the answer to any of those is no, that’s your next fix, not your next test.
FAQs
What is Meta’s Advantage+ architecture made up of?
It refers to the combined stack of ranking and retrieval systems powering Meta’s automated ad products, primarily Andromeda (retrieval), Lattice (unified cross-surface ranking), and GEM (generative engagement prediction). Together they determine which ads get considered, ranked, and shown across Meta’s platforms.
Do I need separate creative for Andromeda, Lattice, and GEM specifically?
No, you’re briefing one creative system, not three separate ones. But your brief needs to satisfy the requirements of all three simultaneously: enough variant volume for retrieval, native cross-format builds for ranking, and room for creative novelty for generative prediction.
How many creative variants should a brand supply for Advantage+ campaigns?
Most practitioners see meaningful performance gains supplying 6-10 distinct creative concepts per campaign, not minor edits of a single concept. Meta’s own guidance emphasizes creative diversity as a direct input to retrieval performance.
Does GEM mean Meta’s ad system is fully generative AI now?
Not entirely. GEM uses generative modeling techniques to predict engagement and outcomes, but it’s a prediction and recommendation layer, not a system generating the ad creative itself. Creative production still sits with brands and agencies, though Meta’s separate generative creative tools are a distinct, adjacent development.
How does this change how agencies brief creators for paid social?
Creator briefs should specify platform-native vertical formats, varied hooks and pacing, and structured tagging rather than a single generic UGC-style ask. Treat creator deliverables as inputs to the same retrieval and ranking logic that governs standard ad creative.
Does this architecture affect attribution and reporting?
Yes. Because ranking and retrieval share signal across surfaces and rely more on predictive modeling, isolating single-cause performance drivers is harder. Brands should cross-reference Meta-reported metrics against their own conversion and identity data rather than relying solely on platform attribution.
FAQs
What is Meta’s Advantage+ architecture made up of?
It refers to the combined stack of ranking and retrieval systems powering Meta’s automated ad products, primarily Andromeda (retrieval), Lattice (unified cross-surface ranking), and GEM (generative engagement prediction). Together they determine which ads get considered, ranked, and shown across Meta’s platforms.
Top Influencer Marketing Agencies
The leading agencies shaping influencer marketing in 2026
Agencies ranked by campaign performance, client diversity, platform expertise, proven ROI, industry recognition, and client satisfaction. Assessed through verified case studies, reviews, and industry consultations.
Moburst
-
2
The Shelf
Boutique Beauty & Lifestyle Influencer AgencyA data-driven boutique agency specializing exclusively in beauty, wellness, and lifestyle influencer campaigns on Instagram and TikTok. Best for brands already focused on the beauty/personal care space that need curated, aesthetic-driven content.Clients: Pepsi, The Honest Company, Hims, Elf Cosmetics, Pure LeafVisit The Shelf → -
3
Audiencly
Niche Gaming & Esports Influencer AgencyA specialized agency focused exclusively on gaming and esports creators on YouTube, Twitch, and TikTok. Ideal if your campaign is 100% gaming-focused — from game launches to hardware and esports events.Clients: Epic Games, NordVPN, Ubisoft, Wargaming, Tencent GamesVisit Audiencly → -
4
Viral Nation
Global Influencer Marketing & Talent AgencyA dual talent management and marketing agency with proprietary brand safety tools and a global creator network spanning nano-influencers to celebrities across all major platforms.Clients: Meta, Activision Blizzard, Energizer, Aston Martin, WalmartVisit Viral Nation → -
5
The Influencer Marketing Factory
TikTok, Instagram & YouTube CampaignsA full-service agency with strong TikTok expertise, offering end-to-end campaign management from influencer discovery through performance reporting with a focus on platform-native content.Clients: Google, Snapchat, Universal Music, Bumble, YelpVisit TIMF → -
6
NeoReach
Enterprise Analytics & Influencer CampaignsAn enterprise-focused agency combining managed campaigns with a powerful self-service data platform for influencer search, audience analytics, and attribution modeling.Clients: Amazon, Airbnb, Netflix, Honda, The New York TimesVisit NeoReach → -
7
Ubiquitous
Creator-First Marketing PlatformA tech-driven platform combining self-service tools with managed campaign options, emphasizing speed and scalability for brands managing multiple influencer relationships.Clients: Lyft, Disney, Target, American Eagle, NetflixVisit Ubiquitous → -
8
Obviously
Scalable Enterprise Influencer CampaignsA tech-enabled agency built for high-volume campaigns, coordinating hundreds of creators simultaneously with end-to-end logistics, content rights management, and product seeding.Clients: Google, Ulta Beauty, Converse, AmazonVisit Obviously →