Thrive Digital

116% growth in AI use. Almost none of it in decisions. Generative AI use in marketing grew 116% in the past year, according to The CMO Survey. Nearly all of that growth sits in production: writing, image generation, and ideation. That's real value, but it's just skimming the surface. The harder, more valuable work is sharper targeting, responsive budget allocation, creative that iterates on actual performance data, and systems that build memory over time instead of starting from zero every quarter. Getting there doesn't mean your team needs to build models or write code. It means understanding what's actually happening inside the platforms you're already running: how automated bidding works, what it's optimizing for, what constraints you can set. Skip that step and you're not really adopting AI. You're outsourcing your decisions to whatever logic the vendor built in. If you don't understand how AI works in your stack, someone else is making those calls for you.

Marketing leaders in 2026 need so many skills just to keep pace. AI now heavily influences how bidding happens, audiences are segmented, and creative is optimized, often without much explanation. Budgets are the first thing cut under pressure. Boards want marketers to have revenue ownership on top of brand and demand gen. There's so much to keep track of. In this month's newsletter, we deep dive into the four skills marketing leaders need today to keep ahead of those pressures and thrive in this rapidly changing environment.

There are two types of campaigns and they have opposite rules. Not knowing them leads to expensive mistakes. Evergreen campaigns are your core, always on channels. Think of them as a combination of inputs: creative, audience, bid strategy, budget. Every one of those inputs can change over time. So underperformance isn't a signal to kill the campaign. It's a signal to swap a component. The question is never "should this exist," it's "what needs to change." Test campaigns work the opposite. These need a hard ship or kill criteria set before launch, not three weeks into the results. Pick your threshold (target CPA, percentage lift, whatever fits the test) before a dollar gets spent. Set it after launch and bias creeps in fast. You either keep funding something you're attached to, or you cut it early because you never believed in it. Confuse the two and you'll kill campaigns that just needed a new audience, and drag out tests that needed to die two weeks ago. If a test is going to fail, make it fail fast.

Cross-channel storytelling isn't about telling the exact same story in the same format on every channel. The concept needs to be taken and shaped for each individual platform and journey stage. Copy-pasting one asset across every platform is a recipe for creative fatigue. Watch Raneesha Peiries explain how to think about cross-channel creative in our latest video.

Cross-channel storytelling sounds simple: tell the same story everywhere. In practice, it means your CTV ad, your Instagram ad, your search ad, and your landing page all need to feel like they came from the same brand, even though a customer might see them in any order, on any device, with zero context from the other ones. On our recent episode of The Hypothesis, our Thrive experts, Raneesha Peiries, Christopher Erickson, and Jorge B. Shaadi Moreno, break down why this is harder than it sounds, and why the real problem usually isn't the creative or the media plan. We also get into: ▪️ Why you can't control the order someone sees your ads in, and what to do instead ▪️ The difference between creative fatigue and creative repetition ▪️ If you can actually prove if cross-channel storytelling worked https://lnkd.in/gKRvgP4b

Campaigns don't just stop working, marketers stop optimizing them. But campaigns and tests are not the same. Evergreen, always-on campaigns can always be improved. New creative, new audience, new bid strategy. Change enough inputs, and you're not killing the campaign, you're running a better version of it. Tests are the real exception. They need clear win or loss criteria set before launch, because once you're mid-test, bias takes over. You either keep feeding a test you're emotionally invested in, or pull one too soon because you doubted it from the start. If it's going to fail, let it fail fast. Want to learn more about how we create tests? Watch our latest Quick Questions with performance expert Chris Ho.

Your AI competitor analysis is only as good as your source data. Most teams skip over that. They pull spend estimates from intelligence tools, run them through AI, and report findings like they're reading a balance sheet. But those numbers are modeled and directional. The tool reps will tell you that themselves. Layering AI inference on top of sampled data doesn't improve accuracy. Watch Director of Strategy, Amanda, talk about what AI does well for competitor analysis in our latest Quick Question.

Creative and performance teams shouldn't work in silos, but they almost always do. These teams have different backgrounds, different priorities, and different definitions of success. Left unmanaged, you get either ads that look great and don't convert, or ads that convert and don't represent the brand. Creative Strategist, Michelle Ng, walks through exactly where the breakdown happens and how to fix it: ▪️ Unclear briefs that leave too much room for interpretation ▪️ No shared feedback loop after creative launches ▪️ A missing north star that neither team is actually aligned on

The best marketing leaders are fluent in finance. When they speak in terms finance already understands (CAC, LTV, contribution margin, payback period), they stop being a cost center asking for money and start being a strategic partner allocating it. That means knowing which product lines actually drive margin, how budget decisions get made, and building one shared view of marketing's financial contribution that finance can stand behind too.

If an ad fatigued, it worked once. That means you have data. Use it. The instinct is to scrap everything and start fresh, but most of your creative output should be iterations on past performers. We often follow the 80/20 rule to leverage what's proven while leaving room to find the next thing. A few things worth knowing as you iterate: 🔶 Meta's Andromeda groups ads with minor differences as the same creative entity. Two ads with slightly different headline copy aren't two ads in the algorithm's eyes. Iterate at the thematic level, not the detail level, or you're capping your own learnings. 🔶 TikTok fatigues faster than Meta. Your refresh cadence should reflect the platform. 🔶 When iterating, isolate what actually drove performance. The angle, the format, the value prop. Find it, then build from it. Your data tells you when to refresh. It also tells you what to keep. Watch Creative Strategist, Michelle Ng, break it all down here.