Across Google, Meta, and other platforms, AI-driven ads promise better performance—but many marketers are discovering they’ve quietly lost the ability to steer. Here’s what’s really changing, and how founders can protect results without clinging to the past.
Loading…
Playbooks on AI and Web3 product marketing, GTM and distribution. No spam, unsubscribe anytime.
Written by Kevan Shah, Web3 and AI product marketing strategist and founder of Hangryfeed.
Head of BD at Zealy and growth operator with 8+ years experience helping AI and Web3 products raise $26M+, scale to $2.2B monthly DeFi volume, and reach 1.2M+ users.
Late on a Thursday night, a marketer opens her Google Ads dashboard and feels her stomach drop.
Headlines she never wrote are now live in search ads. An “auto-generated assets” toggle has quietly flipped itself on. Budgets are flowing into keywords she’s never approved—and some barely relate to her product. When she digs deeper into reports to understand what happened, all she finds are thin summaries and missing details. The system simply tells her, “Performance is improving.”
This is the new tension in digital advertising: AI-driven ad platforms are promising better performance while quietly taking away control.
Over the past few years, the biggest ad platforms have been rolling out AI-first campaign types—Google’s Performance Max and AI Max, Meta’s Advantage+ products, TikTok’s “smart” optimization, and similar tools elsewhere.
The pitch is appealing on the surface:
At the same time, familiar controls have been shrinking or being pushed into obscure corners:
Many marketers are discovering changes only after AI-written copy and expanded targeting go live on real budgets. They’re not being asked, “Do you want this?” The platform is assuming the answer is yes.
Two very different reactions are emerging.
For one group of practitioners, these tools feel less like helpful assistants and more like black boxes with access to your wallet.
Their frustrations cluster around a few themes:
Silent defaults and loss of consent
They report account-level and campaign-level AI features turning on without clear, explicit approval. Auto-generated assets start producing headlines and descriptions; AI expansion begins matching ads to new queries and audiences; placements widen automatically.
The pattern feels less like “try this new feature” and more like “you’re now opted in unless you notice and turn it off.”
Irrelevant traffic, wasted spend, and brand risk
Where older setups allowed granular control over keywords, audiences, and placements, AI-first campaigns often obscure the “why” behind results.
Marketers describe spending extra hours hunting for irrelevant queries, excluding what little they can see, and trying to guess what the algorithm is doing—all while being told the system is “learning.”
Vanity metrics over business outcomes
Another pain point: the algorithm optimizes for whatever metric you define, even if it’s a weak proxy for real value.
If the goal is set to “leads” without quality checks, AI will find more leads—often low-intent, low-quality, or even spam. If the objective is “add to cart,” it will maximize adds to cart, not profit.
The result is campaigns that look great on the platform dashboard while sales teams struggle with junk and founders wonder why revenue hasn’t budged.
Skill erosion and dependency
Many experienced practitioners feel their expertise is being devalued. Years spent mastering bid strategies, keyword structures, and manual optimization are now being replaced by a system that says, “Don’t worry, we’ve got this.”
They worry that:
In this view, AI isn’t just changing the tools—it’s actively boxing marketers into narrower, shallower roles.
The other camp sees the same tools and feels something very different: relief.
For them, automation is not a takeover; it’s a chance to stop living in spreadsheets and start doing real marketing.
Their core arguments are just as strong:
Control is shifting, not disappearing
Yes, they acknowledge that you lose certain knobs:
But they argue that control has moved upstream:
In this view, the human role moves from micromanaging delivery to architecting the system’s inputs.
Performance can improve—sometimes dramatically
This group has real results to back up their optimism:
They often point out that many “AI horror stories” come from weak implementation:
Their message to founders is blunt: if you feed bad data into automation, you get bad outcomes. Fix the inputs before blaming the algorithm.
Marketers are finally freed to focus on strategy
The most hopeful voices see AI as a chance to upgrade the marketer’s job:
They warn against nostalgia for knobs that may not matter as much in a data-rich, AI-driven environment—and urge teams to climb “up the stack” into strategy and systems.
Underneath the feature lists and campaign names, this debate is really about three things:
1. Who defines success?
If the platform decides your goal is leads, sign-ups, or views—and you accept that definition uncritically—you will get more of those numbers, whether or not they move your business.
Founders and marketers need to reclaim this step:
2. Whose incentives are baked into the AI?
Ad platforms make money by serving more ads and increasing auction revenue. Their optimization logic is not neutral; it’s tuned to grow their business.
Automation that looks “helpful” can also:
This doesn’t mean automation is bad. It does mean teams should layer their own measurement and guardrails on top of platform reporting—not accept glossy performance summaries at face value.
3. How much opacity is acceptable?
Every system has complexity, and not every lever can stay exposed forever. But there’s a difference between:
Marketers are wrestling with where that line should be—and what level of transparency is “good enough” to manage risk.
If you’re running campaigns today or planning your next growth push, you don’t have the luxury of sitting out this shift. AI-powered ad products are becoming the default, whether you love them or hate them.
You do, however, have choices in how you engage.
Here are grounded actions that reflect what practitioners are learning at the coalface:
1. Treat automation as a tool, not a strategy
AI Max, PMax, Advantage+, and similar products are tactics. Your strategy still needs to answer:
Start with those questions and then decide where automation helps and where you need manual control.
2. Own the definition of “good” and bake it into your tracking
Do the hard work of defining and instrumenting true success:
If you’re short on data, start simple but honest: focus on signals that clearly correlate with revenue, then add sophistication over time.
3. Use guardrails and hybrid setups instead of going all-in blindly
You don’t have to choose between “full AI” and “full manual.” Many teams are finding balance through:
Think in terms of experiments rather than permanent commitments.
4. Audit defaults regularly
Make it a habit to:
This doesn’t require paranoia—just deliberate ownership of your environment.
5. Invest where humans still have a clear edge
As targeting and bidding become more automated, the most defensible skills shift toward:
These are not easily automated, and they heavily influence what AI can do for you.
You don’t need to fight the algorithm on its turf. Focus on the levers only you can pull.