We’ve been running paid media for brands since 2009. We’ve watched Facebook go from a curiosity to the backbone of most DTC growth strategies. We’ve adapted through iOS 14, the death of the pixel, the rise of Reels, and about a dozen algorithm overhauls that were each described as “the biggest change ever.”
This one actually might be.
Mark Zuckerberg is not being subtle about where this is going.
In a May 2025 interview, he laid it out plainly: “You’re a business, you come to us, you tell us what your objective is, you connect your bank account, you don’t need any creative, you don’t need any targeting demographic, you don’t need any measurement, except to be able to read the results that we spit out.”
He wasn’t describing a distant future. He was describing the direction Meta is actively building toward, today.
The money behind the machine
AI can do a lot. It also costs a lot. Bain’s 2025 Global Technology Report put a number on just how expensive it is: to meet projected AI demand by 2030, the industry needs roughly $2 trillion in annual revenue to fund the required data center build-out. Even after accounting for AI-related efficiency savings, there’s still an $800 billion gap. That “math has to math”. For platforms like Meta, the most direct path is advertising revenue, which means every dollar of investment in AI-driven ad tools needs to justify itself through advertiser adoption and spending.
That pressure is already showing up in how Meta builds its products.
Fewer knobs, less control, more algorithm
Since 2024, Meta has been systematically reducing what advertisers can manually control. Specific interest targeting categories have been consolidated into broad groups or removed entirely. Detailed targeting exclusions were eliminated in early 2025. The entire campaign creation experience has been restructured to default to Advantage+, Meta’s AI-driven campaign framework, with manual options increasingly treated as the exception rather than the rule.
The result is that audience targeting as most advertisers have practiced it is becoming a suggestion, not a mandate. You can still tell Meta roughly who you want to reach. The algorithm decides whether to listen. The new platform, called Andromeda, completed its global rollout in October 2025. Every advertiser on Facebook and Instagram is now running through it, whether they know it or not.
When targeting becomes a black box, creative becomes the targeting
Here’s where it gets practical for brands.
As the manual levers disappear, the creative itself is increasingly the primary signal the algorithm has to work with. If your ad shows a specific person with a specific problem finding a specific solution, the algorithm has something to pattern-match against. It can find more people who look like the ones engaging with that narrative.
If your creative is generic, the algorithm has less to work with. It either casts too wide a net or keeps serving the same small audience until performance collapses.
Meta’s own research has confirmed what every performance team already knows from experience: repeated exposure to the same creative drives conversion rates down and costs up. In today’s environment, fatigue sets in faster than it used to. Andromeda concentrates spend aggressively behind winning creatives and burns through audiences quickly. What used to have a lifecycle of six to eight weeks might now fade in two or three.
The practical implication is that creative volume is no longer a nice-to-have. Brands that can produce, test, and rotate a steady stream of high-quality assets have a structural advantage. Not different sizes of the same ad. Different angles, hooks, proof points, use cases, and formats. You are feeding the machine better data to learn from.
What this means for your brand right now
Paid social isn’t broken. Just the way most brands have run it is getting disrupted. And the disruption is moving fast.
A few things we believe are true heading into this environment:
Creative quality and volume are your primary competitive levers. If you are not thinking about creative as a system rather than a project, you are going to feel this.
Also, first-party data matters more than ever. Clean customer lists, strong pixel events, and tight conversion tracking give the algorithm better signals to work with. Brands that are sloppy about measurement are handing an advantage to their competition.
Platform diversification is risk management. If Meta’s AI decides your account has peaked, you want alternatives that are already working. Search, retail media, email and SMS, affiliates, and creators should not be afterthoughts. They should be part of a growth stack that does not depend on a single platform’s algorithm staying friendly.
The question we are asking for every client right now is not “how do we target better?” It is “how do we give the algorithm better inputs?” Those are very different questions, and they lead to very different decisions.
If you want to think through what this means for your brand specifically, let’s talk.
0 Comments