Meta has spent the last few years automating the core of what most performance marketing teams still consider their job. The targeting architecture, the bid adjustments, the placement decisions, the audience segmentation. The machine is doing them more effectively now, and the gap between where most teams are spending their time and what is actually moving performance is getting wider.
I spent last week at the Meta Performance Marketing Summit, and that gap was the subject of nearly every presentation. The central message was consistent: performance marketing is moving from manual optimization toward AI-orchestrated systems, and the role of the advertiser is evolving accordingly. The organizations that understand this are building for it. Most are not.
The performance engine has been rebuilt from the ground up
At the summit, two systems dominated the technical conversation, and understanding them is the fastest way to understand why the platform behaves so differently now than it did even eighteen months ago.
Meta rolled out a major Lattice update in February, and we wrote about it at the time with our initial thinking on what it meant for advertisers. At the summit, Meta gave the fullest picture yet of how it actually works. Previously, separate models optimized for separate objectives in isolation: one for engagement, another for conversion, another for reach. What Lattice does is allow all of them to learn from shared behavioral data simultaneously. Purchase behavior now improves engagement prediction. Engagement signals improve conversion prediction. The whole system gets smarter because every part of it is learning from everything at once.
The strategic implication is significant. Meta’s systems are increasingly optimizing holistically across the full funnel, while most advertisers are still running siloed campaigns with fragmented KPIs and disconnected creative systems. The platform is learning cross-funnel faster than most organizations are operating cross-functionally. That is not a gap you can close by working harder inside the existing model.
Andromeda goes a layer deeper. Historically, retrieval systems identified eligible ads, and ranking systems determined what to show. With Andromeda, retrieval itself is now AI-personalized. Meta can assess which ads a user is likely interested in before ranking even begins. They backed this with serious infrastructure: 10,000x more compute power added to retrieval systems through Nvidia partnerships. The compute investment makes the ambition clear. This is a foundational rebuild, not an optimization update.
Taken together, what Lattice and Andromeda describe is a platform that is no longer primarily responsive to advertiser inputs. It is making increasingly sophisticated decisions on its own, upstream of anything a media team touches.
That shift has a direct consequence for how media teams should be spending their time. The things that have historically differentiated skilled media buyers, targeting architecture, bid manipulation, audience segmentation, structural complexity, are being automated more effectively than humans can manage them manually. The new differentiators are creative quality, first-party signal quality, conversion data integrity, product feed quality, and measurement sophistication. If your team is spending most of its time on the things Meta has automated, you are optimizing for the wrong layer of the stack.
The three inputs that now determine whether you win
If the platform is making the decisions that media teams used to make, the question becomes: what do you put in front of it? Three inputs came up repeatedly at the summit, and all three are underinvested by most advertisers.
The first is creative. Meta was explicit: stop trying to find the single winning ad and start building systems that continuously generate and evolve creative signals. Their Catalog Product Video format is the clearest proof point, delivering 20% more conversions per dollar and 33% higher incremental conversions in Reels placements. Meta’s generative AI tools can now produce thousands of creative combinations from existing assets with limited manual effort. The operational shift this demands is significant. Creative strategy is no longer about periodic production cycles or hero assets. It is modular, iterative, and signal-driven. The organizations best positioned to win are the ones treating creative as a continuous feed into the AI system, not as a campaign output produced on a brief cycle.
The second input is creator content. This was one of the most commercially aggressive sections of the summit, and the one most likely to disrupt how agencies and brands are currently structured. Meta has rebuilt its Creator Marketplace to integrate directly with custom audiences, Ads Manager, and performance signals. The evaluation criteria for creators has shifted from follower count and engagement metrics toward performance probability, audience overlap, and business outcomes. Partnership Ads are delivering 19% lower CPA, 13% higher CTR, and 71% improvement in brand sentiment when integrated into BAU campaigns. The framing from Meta was unambiguous: creator content is no longer an influencer strategy sitting alongside your paid social activity. It is core performance infrastructure, and the future requires integrated creator and paid social teams, creator scoring systems, scalable sourcing, and incrementality frameworks built specifically for creator programmes.
The third input is product data, and it was the most underrated theme of the summit. Your product catalogue is no longer back-end commerce infrastructure. It is the raw material powering AI-driven personalisation, dynamic creative generation, and contextual commerce experiences. Meta described scenarios where Meta AI recommends products contextually based on behaviour, preferences, saved content, and prior purchases, with upcoming capabilities across top product insights, category benchmarking, brand versus price analysis, and automated product video creation. Feed quality directly determines targeting quality, recommendation quality, and creative quality. Most advertisers are still treating catalogue governance as a technical task. It is a strategic one, and the gap between those who treat it that way and those who do not is going to become visible in results.
Why most advertisers are flying blind on what is actually working
Meta was unusually direct about measurement at the summit, and it was some of the most commercially important territory they covered. Even if you fix your creative, your creator program, and your product data, you are probably still measuring Meta’s contribution incorrectly.
Their data shows that 31% of incremental conversions driven by Meta are being misattributed to other channels. That number represents a structural problem in how most organizations measure performance, with real consequences for budget allocation, channel investment, and strategic decision-making. If you are optimizing to last-click ROAS, you are making decisions based on a picture that systematically undervalues what Meta is doing.
The reason this happens is that much of Meta’s impact lands earlier in the journey: discovery, cultural influence, search behavior shifts, assisted conversions. Users encounter something on Meta and convert elsewhere. Standard platform reporting cannot capture this, and most measurement models are not calibrated to account for it accurately. H&M ran Conversion Lift experiments to calibrate their MMM models and saw a 3x improvement in incremental ROAS across key markets over two years. A gain at that scale is the difference between chronically under-investing in a channel that is working and actually understanding what is driving growth.
Meta’s recommended path involves incrementality testing, experiment-based measurement, Conversion Lift, Brand Lift, MMM calibration, and predicted LTV integration. The harder challenge, which Meta acknowledged directly, is organizational transformation. The tooling exists. What most organizations lack is the structural alignment to use it properly: finance and marketing operating as connected functions, experimentation operationalized rather than occasional, and leadership aligned around incremental business growth rather than attributed click volume. The problems that remain are structural, and most organizations have not yet restructured to solve them.
Bottom line
The summit’s consistent argument was that paid social is becoming systems management, not campaign management. The strategic role of agencies and leaders is shifting toward architecting learning systems, integrating measurement frameworks, operationalizing creative pipelines, improving signal quality, and designing AI-native workflows.
What will separate the organizations that win from those that fall behind is not budget size or platform access. It is the speed at which they can restructure around the model Meta has already built. The technology is largely there. The organizational will to use it is the variable.
