Latest / The CMO Podcast with Fexingo: Marketing Leadership, Budgets, and Executive Strategy / Why CMOs Are Using AI for Dynamic Creative Optimization
Transcript
- Lucas: So Luna, let me ask you something. When was the last time you saw a digital ad that actually felt like it was made for you — not just targeted, but the actual imagery and copy seemed to speak to what you were thinking about? Luna: Honestly, maybe a week ago? I was looking at hiking boots, and an ad from REI showed me a pair I'd viewed, but the headline mentioned a trail I'd just been reading about. I assumed it was personalization on steroids. Lucas: Right, and you're probably right to suspect that. What you experienced is a taste of something called dynamic creative optimization — DCO for short. It's an ai driven approach where the ad itself — the image, the headline, the call to action — is assembled in real time based on who you are. Luna: So it's not just showing the same creative to a segmented audience. It's actually building the creative on the fly. Lucas: Exactly. Traditional programmatic advertising uses machine learning to decide which ad to show you, but the creative is pre-made — maybe ten variants. DCO flips that: the system has a library of components — product shots, background colors, fonts, offers — and it combines them into millions of possible ads, then serves the one that's predicted to perform best for you. Luna: That sounds like a huge leap from A/B testing. What does it mean for a CMO's budget? Are we talking about replacing creative teams? Lucas: Not replacing, but definitely reshaping. I want to talk about a real example — a large retailer I'll call 'National Outdoor Co.' — not their real name, but the numbers are real. They had a typical challenge: they sell thousands of SKUs across apparel, gear, footwear. Their old approach was to shoot maybe 50 hero images per season and run A/B tests on two or three headlines. Luna: Which means most of their inventory never got highlighted in a way that matched what a shopper actually wanted. Lucas: Exactly. So they implemented a DCO platform that integrated with their product feed and first-party customer data. Now, when a user visits their site and then sees a retargeting ad, the system pulls the exact product they viewed, combines it with a background color that machine learning has determined converts best for that user's segment, and writes a headline that includes a relevant benefit — like 'Waterproof and Ready for Rainy Trails' if the user's location is getting precipitation. Luna: That's hyper-personalized. And they measured results? Lucas: They did. Over a three-month campaign, they saw a 35% increase in conversion rate from retargeting ads, and a 50% reduction in creative production costs. They didn't need to shoot a hundred different images — instead, they invested in a modular creative system and an AI platform. Luna: So the savings come from not having to produce so many static variants. But what about the tech side? How does the system learn what works? Lucas: It's reinforcement learning, essentially. Every time an ad is served, the platform collects a signal — click, conversion, time on site — and feeds that back into the model. Over thousands of iterations, it figures out that for a 30-year-old urban female who clicks on trail-running shoes, a minimalist ad with a single product shot and a 'Free Shipping' call to action works better than a lifestyle image with a long headline. Luna: That sounds powerful, but also a little creepy. How do consumers feel? And what about privacy regulations? Lucas: Those are the key concerns. The DCO systems rely on data — and in a privacy-first world, that data needs to be first-party or sourced through consent. The retailer I mentioned uses their own loyalty program data and site behavior, not third-party cookies. They also allow users to opt out of personalization entirely. The ad still shows, but it's a generic version. Luna: So the onus is on the brand to be transparent. I can see how this becomes a CMO priority — it's not just a tech play, it's a brand trust play. Lucas: Exactly. And that's why we're seeing more CMOs appoint a 'personalization lead' who sits between the creative and data teams. The technology itself is becoming commoditized — there are a half dozen DCO platforms now — but the real competitive advantage comes from the quality of your data and the creativity of your modular assets. Luna: Let's talk about that creative side. If you're a CMO, how do you structure your creative team to feed a DCO system? Lucas: It's a shift from campaign-based thinking to always-on asset libraries. Instead of producing a set of ads for a three-month push, you produce components — 20 product shots, 10 background textures, 15 headline templates, 5 call to action buttons. The system does the rest. You also need to define rules — for example, never pair a luxury product with a discount headline, or always include the brand logo in the top-left corner. Luna: So the creative team becomes more like architects of a system rather than authors of individual ads. Lucas: Right, and that requires a different skill set. Some agencies are now offering 'dco ready creative' as a service. They'll produce a library of assets that are designed to be mixed and matched by an AI. Luna: What about measurement? How do you know if DCO is actually working compared to a traditional approach? Lucas: You run a holdout test. You take a segment of your audience — say 10% — and serve them only static, non-dynamic ads. Then compare the performance. National Outdoor Co. did that and saw a 22% lift in revenue per user from the DCO group. That's tangible. Luna: Twenty-two percent is huge. But does it work for every industry? I imagine it's easier for e-commerce with lots of SKUs, but what about B2B or services? Lucas: It works best when you have a large product catalog or multiple value propositions. But B2B companies are using it too — swapping out case study images, industry-specific headlines. For example, a software company might show a healthcare customer a testimonial from a hospital, while a manufacturer sees a factory floor photo. The logic is the same. Luna: And the cost? Is this something only enterprise brands can afford? Lucas: The platforms have gotten more accessible. A few years ago, you needed a minimum spend of six figures. Now there are self-serve options starting at a few thousand dollars a month. The bigger investment is the creative production and the data infrastructure. But for many brands, the ROI justifies it within a quarter. Luna: So any CMO listening should at least be running a pilot. What's a good first step? Lucas: Start with your highest-traffic retargeting campaign. Use the data you already have — product views, cart abandons. Create three modular assets: a product image, a background, and two headlines. Test DCO against your current static approach for two weeks. You'll learn fast. Luna: And if it works, you scale. If not, you've only invested a small budget. Lucas: Exactly. And that's really what we're about on this show — giving CMOs practical tools they can implement. Speaking of which, if these conversations have sparked something you've actually used in your own marketing, we'd love to know. The show is ad-free, and that's thanks to listeners who support us directly. If today's episode was worth a coffee to you, you can find us at buy me a coffee dot com slash fexingo. No obligation, just a way to keep these insights coming. Luna: It's a small gesture that makes a big difference. And we genuinely appreciate it. Lucas: Now, back to DCO. One thing I want to emphasize is that this isn't a set-it-and-forget-it tool. You need to monitor for brand safety — make sure the AI isn't combining assets in ways that are off-brand or offensive. Human oversight is still crucial. Luna: So the CMO's role becomes one of governance and direction. Lucas: Exactly. And that's the exciting part — the technology handles the heavy lifting of personalization at scale, but the strategy and the creative vision still come from people. DCO is a tool, not a replacement. Luna: Well, I know I'll be looking at ads a little differently now. Maybe I'll even click one on purpose to see what happens. Lucas: Just be careful — you might end up buying those hiking boots. I know I'm tempted.