Why AI in Performance Marketing Still Needs Great Human Input
TL;DR
AI can quickly generate landing pages and ad hypotheses, but it's not the main factor behind achieving results. Cody Plofker, CMO of Jones Road Beauty, uses Claude to transform customer reviews, support tickets, and heatmaps into functional pages in a single session, though he remains actively involved throughout the process. His view is that AI broadens the scope of what an individual can produce, but it doesn't replace the critical judgment needed to determine if a test is genuinely worthwhile. He's also skeptical of AI-generated user-generated content, prefers to treat ads and landing pages as a unified project rather than separate tasks, and closely ties team KPIs to output to encourage genuine adoption.
Who This Is For
Media buyers and creative strategists running high-volume testing who want to speed up production without losing quality control. Especially anyone building landing pages and ad creative separately and feeling the friction of it.
Want the AI workflows before everyone else is copying them? Subscribe to AI DTC WTF and get the strategies straight from operators actually running the tests.
In a recent AI DTC WTF podcast episode, Cody Plofker, CMO at Jones Road Beauty shares how he’s turned AI (specifically Claude) into a full-stack CRO and conversion engine. We’ll dig into his workflow, why human taste still matters, and the new standards for output in the age of AI.
Cody’s End-to-End CRO Workflow
Cody always starts with data before starting an AI prompt. He has a running “customer intelligence” document where it constantly pulls in live customer reviews, support tickets, survey responses, GA4, heatmaps, and more. Then he feeds all of that into Claude, and suddenly he’s got hypotheses, wireframes, even full landing pages ready to go.
He has now built out a whole brand design system inside a GitHub Markdown file, so then he can spin up a new landing page at about 80% quality in just one session. That’s a serious unlock.
“The strategy is more important than the execution," says Cody. "With AI, you can run more tests, but if your tests aren’t good, you’re just moving faster in the wrong direction."
Why Human Taste Still Matters
AI can crank out ideas faster than ever, but Cody’s super clear on this: it’s not perfect, and it’s definitely not going to replace human judgment. He’s hands-on at every stage, from briefing to reviewing to refining, making sure every test and every page actually lines up with what customers want and what the brand stands for.
“You can train AI on all your frameworks, but you still need someone with taste and context in the loop. The final 20% of quality always takes the most time.” - Cody Plofker
Ads and Landing Pages: Build Together, Not in Silos
Cody makes it clear that he doesn’t treat ads and landing pages like separate silos. Instead of building a landing page and then scrambling to find ads that fit (or launching ads then scrambling to match those to a landing page), his team starts with one brief and builds both together. That way, they can move fast, test more, and sometimes launch a whole new funnel in under a day.
The Limits of AI UGC (And the Real Role of Automation)
Cody’s all-in on AI for operations, analysis, and workflow, but not for AI-generated UGC. He says: “I’m pretty bearish on AI content. Static ads are one thing, but with AI UGC, you’re so far off. The real value is automating the backend so you can spend more time being creative and talking to customers."
He sees the future of creative strategy as less about wrestling with spreadsheets and more about really digging into customer angles, while AI just brings the right data and trends to the surface automatically.
If AI UGC is a trap, this is the way out. Browse real creators built for DTC brands at The Hive Haus.
Getting Teams to Actually Adopt AI
Cody says that even with all the training and incentives to fully adopt AI in workflows, employees and contractors will always tend to fall back on their old habits. The real shift happens when there’s some urgency, like tight resources or a clear push from leadership. And as Will Sartorius points out, leaders have to be upfront about why this matters, show what’s possible, and make sure there are real incentives tied to adoption. These incentives include KPIs and performance metrics so that there's trackable goals and progress. Cody says he’s now linking KPIs (and even bonuses) directly to what employees ship: the number of ads launched, funnels built, and speed from idea to execution. He says that AI isn’t just a nice-to-have and it's become as essential as Slack or email.
Key Takeaways
Data-driven creative is non-negotiable: Pull insights from customer reviews, support tickets, surveys, and heatmaps to guide CRO and creative.
Automation is for operations, not inspiration: Use AI to handle the grunt work so humans can focus on strategy and creativity.
Ads and landing pages should be briefed and built together: This enables faster iteration and tighter alignment.
Human-in-the-loop is essential: AI can accelerate, but not replace, good taste and judgment.
Mandate AI adoption, but explain why: Incentives, transparency, and urgency drive real behavior change.
Raising the bar: AI lifts the ceiling on what a single operator can do, but expectations for everyone are higher than ever.
This blog post is inspired by the AI DTC WTF podcast episode “Why AI UGC Is a Trap (And What Actually Works) | Cody Plofker”. Check out the full episode below.
Note: This blog was written by a human, but AI helped draft the outline and synthesize ideas.
