Stop trying to out-calculate the machine and start feeding the machine better signals was the theme from Ginny Marvin, Google’s Ads Product Liaison, during a recent episode of the Ads Decoded podcast she hosts. To many, it sounded like a victory lap for automation and seemed to set the industry on fire. To others, it felt like a final surrender of the steering wheel.
We are currently navigating a mass handover of campaign control to automated systems, and the speed of this transition is frequently outpacing our understanding of what we are surrendering. The numbers confirm that this isn’t just a trend; it is the new baseline for performance marketing. More than 1 million advertisers have now adopted Google’s Performance Max globally. On Meta, Advantage+ campaigns now account for 35% of all U.S. retail ad spend. Even TikTok has seen its Smart+ automated solutions jump from a mere 9% to 42% of performance campaigns in a single year.
The platform narrative is seductive. Google recently rolled out new steering and reporting updates for Performance Max, including audience exclusions and budget reporting, to address the long-standing “black box” criticism. According to Meta’s own engineering data, advertisers who adopted Advantage+ creative features saw an average 22% increase in return on ad spend, although results vary significantly based on first-party data quality and campaign maturity. But there is a dangerous gap between these platform claims and real-world performance that every SEO and paid media specialist needs to acknowledge.
A new report from Adtaxi hits the nail on the head: AI does not replace strategy; it magnifies it. If you provide the algorithm with strong data inputs and a clear definition of business value, then you get powerful outcomes. If you provide weak inputs, then you simply produce “accelerated inefficiency.” The machine will spend your budget with incredible speed, but it cannot navigate the strategic complexity that exists outside its training data.
In the era of GEO and entity-based search, the discipline required to feed ad platforms accurate, high-quality signals is the same discipline that builds brand authority in organic and AI-driven search results. When we talk about “the machine,” we are really talking about an interconnected ecosystem of data. If your ad campaigns are optimizing for surface-level metrics rather than true business outcomes, then you are essentially training the platforms to misunderstand your most valuable customers. If your SEO campaigns don’t include the prompt topics that your target audience is using, then read this.
For instance, Google’s latest April 2026 updates for Performance Max allow for first-party audience exclusions. This sounds like a technical setting, but it is actually a strategic pivot. It allows marketers to stop wasting acquisition budget on existing customers and focus on true growth. However, this exclusion is only as good as the CRM data behind it. If your first-party data is messy, your “automated” efficiency is an illusion.
We see this in the attribution gap on platforms like TikTok, where traditional last-click models fail to capture up to 79% of the conversions that automated systems are actually driving. Without a human expert to validate and measure these systems against real-world goals, we are just watching the algorithm spend money in a vacuum.
I contacted Jennifer Flanagan, vice president of Marketing at Adtaxi by email, and she countered that the lack of transparency in these systems creates a genuine risk where systems optimize for platform-defined metrics rather than business health. She correctly identified human experts as the “steadying hand” of strategy that machine learning cannot replicate.
The Lesson For 2026
It’s a clear lesson that you cannot “set and forget” your way to market leadership. The most successful marketers follow a strict rule of resource allocation: Invest the vast majority of your energy into human talent and strategy, and let the remaining fraction go toward the tools themselves. AI is running more of your advertising than you probably realize. The only question that matters now is whether you are running the AI, or if you are simply watching it spend your budget.
More Resources:
Featured Image: Master1305/Shutterstock
