Meta Performance Marketing Summit 2026: What You Need to Know
Why it Matters
Meta just sent a clear message to advertisers: the old ways of doing things are becoming outdated. If you're still measuring success with last-click attribution, focusing only on the very end of the customer journey, and treating creators as optional partners, you’re working with a version of the platform that’s on its way out.
This isn’t just a small change; it’s a significant shift. Meta is redefining how ads are ranked—now it’s about the entire sequence of interactions, not just those final moments. They’re also shifting the way we should measure performance; it’s no longer just about clicks but about the overall lift in engagement. The algorithm is rewarding richer context, diverse creative approaches, and genuine signals from creators. These aren’t just tweaks; they’re foundational changes.
The brands that will really thrive are the ones that start adapting now, providing the system with better data and insights before their competitors catch on. Companies that wait until these trends are obvious will find themselves playing catch-up against those who’ve already established a lead with valuable data.
Meta's fifth-annual Performance Marketing Summit in San Jose laid out a clear picture of where the platform is heading for 2026 and beyond. The core messages were clear:
AI is the operating system and the foundation of Meta ads. It's no longer a nice-to-have or a learn-it-when-you-can entity.
Last-click attribution is being retired in favor of incremental attribution.
New customer acquisition is the next major performance lever Meta wants advertisers to pull on.
The algorithm stack (Lattice, Andromeda, GEM) is moving toward sequence-aware, person-level personalization and funnel consolidation.
Creators are the single most efficient input into the entire Meta ranking system.
If you missed the event in San Jose (or were there and need a refresher), this is the recap that will make you feel like you were on the front row digesting everything Meta had to say with the best performance marketers in the country.
The mainstage day at the Meta Marketing Summit in San Jose on May 13, 2026 was packed with product announcements, algorithm updates, and a strong point of view from Meta on where performance advertising is going. Five themes carried through almost every session: AI everything, incremental attribution, new customer acquisition, sequence learning across the algorithm stack, and creators as the new (and also not-so-new) performance powerhouse.
AI is now the throughline of every Meta product
The summit opened with the assumption that AI is no longer an emerging area for Meta, it is the foundation. Business AI agents were referenced repeatedly as the next major shift, with the platform building toward agents that can help advertisers with setup, optimization, and even creative direction inside the Ads Manager workflow. The framing was less about a single AI feature and more about every Meta ads surface getting an AI enhancement underneath it and AI being the driving factor behind everything Meta ads - audience targeting, placement choices, creative understanding, and more.
A few practical implications came out of the AI sessions, such as the fact that the platform is asking advertisers to provide significantly more context to the system than ever before. Meta's models are now trained to consume not just performance data, but the reasoning behind why something worked or did not work. The clear ask is that advertisers constantly and consistently feed information back to the platform: what success looks like for the business, what failure looks like, what margins matter, and what counts as a genuine conversion. The more specific the input, the better the prediction.
The other AI message was about pace. Meta openly acknowledged that the speed of model development is so fast that no advertiser can stay current on their own. One discussion suggested the approach of having internal study groups within your business and pair team members into AI buddies for weekly learning, treating AI literacy as a required skill across the team rather than something owned by one specialist. The underlying message for performance buyers is that AI fluency is now table stakes. The accounts that consistently feed Meta high-context signal are the ones that will compound a noticeable performance advantage over the next several months.
Sequence learning and the next phase of the algorithm stack
The most technically dense session of the day went deep on the algorithm models powering Meta's ranker. Most advertisers already know Andromeda. Many have heard of Lattice and GEM. Fewer are familiar with the term Adaptive Ranking Model (ARM), but it is going to come up constantly over the next year.
Here is how the four pieces fit together as Meta described them.
Lattice was the first major step, and it allowed Meta to bring multiple machine learning models together into a unified architecture. The Q1 2026 Lattice update is already showing deep-funnel performance increases of 13% CTR lift and 16% CVR lift on the accounts Meta tested.
Andromeda sits on top of Lattice and is focused on the personalized ad retrieval stage. Meta built custom hardware to bring 10,000 times more compute to retrieval, which is what allows the system to read your creative (image pixels, video visuals, and audio cues) and personalize delivery for each individual user at the ranking stage. Andromeda is the reason creative now does the targeting.
GEM (Generative Ad Recommendation Model) is the layer that uses LLM-style training on ad sequences. GEM is what makes Meta capable of asking the more important question: not just whether an ad is relevant to a user, but whether it is the right next ad to show that user given everything they have already seen and interacted with.
ARM (Adaptive Ranking Model) is the newest addition and the one most worth understanding now. Some users have huge interaction histories on Meta, they click on ads constantly and spend significantly more than the average. The previous models truncated those long histories, which limited how well the system could predict the next best ad for high-value users. ARM removes that ceiling and is purpose-built for users who are most likely to click and convert. Meta reported a 3% CVR increase and a 5% higher CTR with ARM in early testing, and it focuses heavily on timeliness, getting the right ad in front of the right person at the exact right moment in their conversion journey.
The reason this matters for performance buyers is that the entire stack is moving in one direction: sequence-aware, person-level, full-funnel ad delivery. Mastering visual format diversity inside Andromeda is no longer the finish line, it is the entry ticket. The next layer of advantage comes from feeding the system enough creative variety to support every point in a user's sequence, not just the bottom-of-funnel conversion ad. The brands that show up at the awareness, consideration, and purchase moments inside the same person's feed are the ones the new stack will reward.
Incremental attribution is the new measurement standard
If there was a single non-AI message Meta wanted advertisers to walk away with, it was that last-click attribution is no longer fit for purpose. Incremental attribution (IA) was framed as the way forward, and Meta brought receipts.
Two stats stood out. On a large-scale test using a seven-day click window, Meta purchases were underreporting by roughly 15%. Roughly one-third of the actual marketing impact in the test was happening downstream on Amazon and other retail channels, which last-click attribution simply cannot see. Just as importantly, 64% of conversions attributed through incremental attribution were new customers, a number traditional attribution tends to badly underweight.
Meta also walked through a four-step path to incrementality: assess, experiment, calibrate, operationalize. The order matters. Most brands run the experiment, see results that conflict with their internal source of truth, and stall out. Meta's point was that running a single conversion lift study is a snapshot in time, not a methodology. The advertisers gaining real ground are the ones using incrementality to calibrate other models (running conversion lift to calibrate MTA or MMM, for example), running tests on a regular cadence, and bringing finance and leadership into the conversation from the start.
Incrementality is a team sport. You cannot adopt it as a buyer alone and expect the rest of the org to come along. Organizational buy-in across finance, leadership, and analytics is the prerequisite for actually changing how a brand makes spend decisions based on lift instead of attribution.
The most actionable line from the IA sessions was simple: just test IA. Even if the first set of results does not match what you are used to seeing, keep testing. The brands that get incrementality-led measurement working ahead of competitors will have a durable advantage that compounds over time.
New customer acquisition is the next major performance lever
Meta is sharpening its focus on new customer acquisition (NCA), and the case is hard to argue with. Only 5% of any brand's total addressable market is currently in-market at any given time. The other 95% are either future buyers or out of the buying cycle entirely. If a brand is optimizing only for the in-market 5%, the algorithm is being told to chase a tiny pool and ignore the much larger one.
This is also where the incrementality and NCA stories converge. The fact that 64% of IA-attributed conversions were new customers reinforces the same point: when measurement only counts last-click, brands tend to overspend on retargeting and undercredit the campaigns that are actually growing the business.
Expect continued investment from Meta in tooling that makes NCA easier to operationalize. Customer Lifecycle Optimization (covered in earlier Foxwell posts) is one piece. Value optimization improvements and tighter exclusion accuracy at the ad set level are others. The throughline is that Meta wants advertisers to be able to explicitly tell the system to go find net-new buyers and trust that the system will do exactly that. Meta also referenced a new solution coming to make NCA and IA work more powerfully together, which is worth keeping an eye on as more details surface.
Creators and the Creative-first approach to Meta Ads
Creators were arguably the loudest single message at the summit. Multiple sessions returned to the same point: creators are the most efficient unit of input into the Meta ad system.
The supporting data is striking. 76% of Gen Z are open to hearing about brands from a content creator. More than three in four social media users prefer creator content over traditional ads. People surveyed are 50% more likely to purchase an item that has been promoted by a creator on Reels. Inside an Andromeda-led ranker that reads pixels, audio, and visual signal to determine relevance, creator content carries more credible signal than almost any other ad type.
Partnership ads were positioned as a core business strategy, not a nice-to-have. The recommended workflow Meta has been hammering for two years now still applies: analyze, identify, launch, expand. Sourcing creators should run on two tracks at the same time: agency partners for high-touch work and the rebuilt Meta Creator Marketplace hub for scale. The two are not in competition with each other, they serve different parts of the same operation. Hit rate matters more than volume, and the smart move is to identify creators who are already statistically likely to perform for the brand before going wide.
A few practical pieces of guidance came up consistently. Embrace creative autonomy and let creators do what they do best on Meta's surfaces, then let Andromeda handle the matching and distribution. Pair partnership ads with ASC campaigns, this combination was repeatedly described as the most reliable performance pattern on the platform right now. Pair ASC catalogs with UGC where possible, and start small to make sure tech connections (product feed, catalog, pixel) are clean before scaling. Break down the silos between creative, media, and analytics teams so that clear comms and aligned KPIs let the process move faster without bottlenecks.
The shift in commerce is reinforcing all of this. The funnel is collapsing. People now go from first discovery to conversion inside a single creator asset, often UGC. Amazon influencers can already tag brand products directly inside Facebook video (with Instagram functionality coming soon), and Amazon is boosting creator content on Meta as part of its own social commerce push. Roughly 80% of consumers now rely on AI shopping or zero-click results for at least 40% of their searches, 92% use social and discovery platforms for product information, and 80% report purchasing after seeing video content. The Meta-to-creative pipeline is no longer optional, it is the engine that feeds the entire ranker.
What's coming to Meta soon
A handful of features were previewed for release over the coming months. Each of these is worth watching closely.
Tagging products in Creator reels and partnership ads (via Shopify/your ecommerce platform AND Amazon products)
Business AI agents
Adaptive ranking model (ARM)
Video catalogs (Focusing on a video-first consumption model)
Improvement and even more focus on Value optimization
Customer Lifecycle Optimization
A rebuilt Creator Marketplace hub designed for easier use and for bridging the gap between organic and paid
Facebook Collabs coming to the hub, opening up a new pool of creators
This blog post is nerdy, but the information we’re discussing (this topic included) is happening DAILY in the Foxwell Founders membership. If you’re interested in this topic, you’re going to LOVE the collaboration and community we have with over 550 members across the world, all expert marketers in their fields of Meta ads, creative strategy, Google ads, Email, CRO, brand ownership, agency operations and more.