- Tension: Marketers celebrate AI-driven efficiency gains without examining whether the original benchmarks deserved to exist at all.
- Noise: Percentage-based improvements create an illusion of progress that discourages deeper questions about strategic foundations.
- Direct Message: Efficiency without clarity of purpose accelerates you toward the wrong destination faster than ever before.
To learn more about our editorial approach, explore The Direct Message methodology.
Most people get this wrong: they assume that a percentage improvement means something moved in the right direction.
A 20% jump in ad efficiency. A 35% reduction in cost per acquisition. A 15% lift in engagement. These numbers land in quarterly reports and boardroom decks like small victories, each one confirming the narrative that the machine is working, that the investment is paying off, that the strategy is sound.
But here is the question almost nobody pauses long enough to ask: efficient compared to what?
If your baseline was a bloated, poorly targeted campaign built on outdated assumptions about your audience, then a 20% improvement means you are now slightly less lost. The compass is still broken. You are walking faster in the wrong direction.
I keep a journal of marketing campaigns that failed spectacularly. I call it my “anti-playbook.” And what I’ve noticed, flipping through years of entries, is that the most dangerous failures rarely look like failures at the moment they happen. They look like incremental wins. They look like efficiency metrics trending upward on a dashboard while the underlying strategy quietly rots from the inside.
The Comfortable Illusion of Better Numbers
Consider what happened when Coca-Cola reported a 20% increase in ad efficiency after leveraging AI to optimize its marketing campaigns. On paper, this is a compelling data point. It suggests that artificial intelligence can refine targeting, reduce waste, and produce more return per dollar spent. And that may very well be true.
But the figure invites a follow-up that rarely gets asked in the rooms where these numbers are celebrated: what was the quality of the strategy that preceded it?
This is the tension at the heart of the AI efficiency conversation. Organizations are layering sophisticated optimization tools on top of strategic foundations they have never properly stress-tested. The AI finds patterns in existing data, amplifies what appears to work, and suppresses what does not.
But “what appears to work” is only as meaningful as the framework it operates within. If your campaign was designed around flawed customer personas, outdated value propositions, or channel assumptions inherited from a different era, then AI optimization is essentially polishing a framework that needed to be rebuilt.
During my time working with tech companies as a growth strategist, I watched this pattern repeat across product launches and market expansions. Teams would deploy increasingly powerful analytics tools to optimize campaigns that were strategically misaligned from the start. The numbers improved. The dashboards turned green.
And yet customer lifetime value would plateau, brand perception would stagnate, and the organization would find itself spending more to acquire customers who stayed for less time. The efficiency gains masked a deeper stagnation.
This is a behavioral psychology phenomenon that operates at the organizational level: anchoring bias applied to performance metrics. Once a team anchors to a baseline number, any improvement feels like validation. The 20% gain becomes the story. The story becomes the strategy. And the original question of whether you were measuring the right thing in the first place disappears beneath a wave of positive-looking data.
When Every Voice in the Room Agrees, Something Is Missing
The current conversation around AI in advertising suffers from a specific kind of distortion: the conflation of operational efficiency with strategic effectiveness. These are different things. They require different questions.
And the chorus of voices celebrating AI-driven marketing improvements tends to blur the line between them in ways that serve vendor narratives more than they serve the organizations making purchasing decisions.
Pratik Thakar, Global Vice President and Head of Generative AI at Coca-Cola, put it bluntly at Brandweek 2025: “The genie’s out of the bottle. You either lead with it, or you keep crying about it.” There is a bracing honesty in that framing. AI adoption in marketing is not optional for organizations that want to remain competitive. But “leading with it” and “leading wisely with it” are two very different postures, and urgency has a way of collapsing that distinction.
Meanwhile, a study published in the Journal of the Academy of Marketing Science found that firms focusing on AI in their operations experienced improvements in net profitability, operating efficiency, and return on marketing-related investment while reducing advertising spend.
These findings are significant and worth taking seriously. Yet they also raise a question that the study’s design cannot fully answer: were the pre-AI strategies these firms were running already well-conceived, or did AI simply impose a discipline that the organizations themselves had failed to maintain?
What I’ve found analyzing consumer behavior data sets over the past several years is that the companies seeing the most durable returns from AI are those that did the difficult strategic work before deploying the technology. They clarified their positioning. They validated their audience assumptions with primary research. They built measurement frameworks that captured long-term brand health alongside short-term conversion metrics. Then they handed those foundations to AI for optimization. The tool had something worth optimizing.
The companies that struggle are those who expected AI to be the strategy rather than a force multiplier for an existing one.
The Question That Changes the Equation
Before asking how much more efficient your marketing has become, ask whether the thing you were doing efficiently was worth doing at all. AI will optimize whatever you give it. The responsibility for giving it something meaningful remains entirely human.
This is the insight that percentage improvements cannot capture. Efficiency is a measure of execution. Effectiveness is a measure of intent. The organizations that will thrive in an AI-saturated marketing landscape are those that hold both in tension, refusing to let the speed and precision of the tool substitute for the slower, harder work of strategic clarity.
Building a Foundation Worth Optimizing
Jeremy Goldman, eMarketer’s Senior Director of Briefings, offered a useful observation about Meta’s AI-driven approach: “After a few years of existential hand-wringing, the company has found its rhythm again by doing what it does best: scaling attention and monetizing it with ruthless efficiency.” That phrase, “ruthless efficiency,” is revealing. It describes a capability, and an impressive one.
But efficiency always serves whatever objective sits beneath it. If the objective is clear and aligned with genuine value creation, ruthless efficiency is a competitive advantage. If the objective is murky or misaligned, ruthless efficiency becomes a way to reach the wrong outcome faster than your competitors.
I learned the hard way, years before I started writing, that data without empathy creates products nobody wants. The same principle applies to AI-optimized marketing. The data will tell you what performs. It will not tell you what matters. It will show you which ad variant drives more clicks. It will not show you whether those clicks represent people whose lives are genuinely improved by what you are selling, or people who were momentarily manipulated into an action they will later regret.
The California tech ecosystem, where I’ve spent the bulk of my career observing these dynamics, is full of brilliant optimization engines running on top of strategies that never asked the fundamental question: who are we actually helping, and how?
So before you celebrate the next efficiency metric that lands on your desk, try an exercise. Work backward. Trace the 20% improvement to its origin. Ask what the original campaign was designed to accomplish. Ask whether that objective still reflects your understanding of your customer. Ask whether the measurement framework captures what you actually care about or what was easiest to track. And ask whether a human being, looking at the full picture, would call this progress or simply motion.
AI will continue to transform marketing operations. That transformation is real and, in many cases, profoundly valuable. But the organizations that extract the most lasting value will be those that refuse to let the tool’s sophistication become an excuse for strategic laziness. A 20% jump sounds impressive. Make sure you know what it replaced before you build your next quarter around it.
