In a largely creative industry like advertising, performance marketing can seem a bit soulless. Algorithms, tracking and optimization can help us reach targets quicker and more efficiently. But the creative used in performance campaigns can often look or feel … off.
When there are metrics to hit and placements to fill, creative often takes a back seat. And the sheer volume and speed needed to produce most performance marketing campaigns tends to limit what can be done creatively. So it makes sense to enlist AI tools to supercharge asset production.
But supercharging already mediocre ads doesn’t necessarily mean better engagement or more brand affinity. We can’t simply trust AI to make up for bland creative with sheer volume.
The real performance shift happens when we stop asking what AI can replace and start asking how technology, crafted with a human touch, can push creativity further in a largely non-creative space.
Volume doesn’t fix bad creative
When working with bigger brands, budgets and timelines support human touch. A CPG brand launching a national campaign might spend months workshopping a single spot, testing every frame for cultural resonance.
But when those same brands shift to performance marketing – slicing budgets across dozens of audience segments, platforms and creative variants – the guardrails disappear.
The result is a sea of sameness: stock imagery, generic copy and creative that looks algorithmically assembled. A skin care brand might test 50 ad variations in a week, optimizing for click-through rate without anyone asking whether the imagery reinforces outdated beauty standards or resonates with its Gen Z audience.
This is where the human and machine crossover gets interesting.
AI can make nuance scalable
Instead of looking for efficiency in volume and speed, AI can now layer on culture and taste.
If you need to show a dog running after your half-off tennis balls across 12 markets, you may need an alternate image for Malaysia, where street dogs carry different cultural associations than the golden retrievers dominating Western pet advertising.
This is the kind of nuance that, until now, only a human could catch. And that human likely wouldn’t catch it in a performance marketing campaign running across hundreds of variants.
Traditional performance marketing relies on lookalike audiences and behavioral signals. AI can go deeper, identifying micro-communities and cultural moments that a brand might never surface through standard targeting.
A DTC furniture brand might use AI to discover an emerging design aesthetic on niche subreddits. Or it might identify that the brand’s product is unexpectedly resonating with remote workers in secondary cities, not just coastal urbanites.
Let AI surface tensions, not resolve them
But AI can’t make judgment calls about brand voice, positioning and long-term equity. It can’t tell you whether leaning into a trending meme dilutes your premium positioning.
Performance marketers operate under intense pressure to hit ROAS and CAC targets, often on weekly or monthly cycles. That pressure creates tunnel vision. When the algorithm says ads featuring aggressive discount language perform 23% better, it’s hard to argue with the data. But what if those same ads are training your audience to ignore anything that isn’t 40% off?
This is where human judgment becomes critical. AI should surface these tensions. It can show you that brand search volume drops when you overindex on promotional messaging. It can flag when creative variants optimized for clicks underperform on brand recall or sentiment. But the decision to sacrifice short-term efficiency for long-term health requires human leadership.
Optimize for return customers
The best use of AI in performance marketing isn’t optimizing for the metrics we’ve always tracked. It’s helping us track better metrics.
Last-click attribution has always been reductive, but in an AI-driven environment, it’s actively misleading. If your AI optimizes creative and audience targeting based purely on conversions, you’re training it to find people already closest to buying – not to build relationships with people who might become valuable customers over time.
Smarter brands are layering in signals like brand search lift, repeat engagement rates and sentiment analysis from creative testing. They’re asking AI to optimize not just for the click but also for return shoppers. That might mean valuing a video view that drives a brand search three days later over a click that bounces in five seconds.
An athletic apparel brand might discover that performance ads featuring real customer stories drive lower immediate ROAS than product-focused ads but generate significantly higher lifetime value and organic social sharing. That’s a trade worth making – but only if you’re measuring it.
Performance marketing can be a craft again
Brands no longer have to sacrifice human touch or cultural nuance because of scale. The opportunity isn’t to replace performance marketers with algorithms. It’s to give marketers tools that make taste and cultural fluency scalable.
This is our opportunity to take back performance marketing as a craft and not an algorithm. To build systems where AI handles the volume and speed we’ve always needed, while humans set the creative bar, define what brand integrity looks like and make the hard calls.
Will we use AI to make more of the same mediocre ads faster, or will we use it to finally bring the creativity, cultural nuance and long-term thinking that performance marketing has always lacked?
The brands that figure this out won’t just hit their ROAS targets; they’ll build audiences who actually want to hear from them.
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Follow Tower 28 and AdExchanger on LinkedIn.
