The Creative Is Fine. Your Thinking Isn’t.

Not everything that looks like a win… is one. And not everything that looks broken… actually is.

This week, we zoom out to rethink how DTC brands measure success, adapt to market shifts, and rebuild creative systems that can scale—even when the rules keep changing.

Inside:

• What the tariff “pause” really means (and why it’s not a full reset)

• A different way to think about your paid media measurement and attribution

• A key method to work with AI and ship native creative that actually performs

• Quick Hits 👉 If your Meta performance suddenly tanked this week, find out why here.

In a noisy environment, clarity is the edge.

🧾 DTC Times — May 19, 2025

🌍 MACRO: The Tariff Truce (Sort of)

This past week, the U.S. and China hit pause on their escalating tariff war. The headlines were optimistic: “145% Tariffs Cut to 30%!”

But read the fine print, and a more complicated picture emerges:

• It’s a 90-day truce, not a permanent rollback (White & Case)

• The de minimis exemption is still dead—small parcel imports from China continue to face 120%+ duties (EightX breakdown)

• Importers are panic-buying, spiking ocean freight costs by up to 50% as they rush to land inventory before August (ICIS)

📉 Flexport’s Ryan Petersen noted that ocean bookings are up 275% week-over-week, and warned “there won’t be enough ships” for the rush (source).

📦 This short-term surge masks the longer-term reality: tariff pressure isn’t gone. It’s simply delayed. The underlying trade instability continues, and ecommerce businesses built on low-cost, direct-from-China fulfillment are still at massive risk.

🧠 One smart take comes from Johnny Celario:

Takeaway:

If you’re importing from China, this is your 90-day window to assess and adapt. Reevaluate your freight, fulfillment, and inventory strategies now, not in August.

📊 TRENDS: What is Marginal ROAS & Why It Matters

In a special episode of Chew On This, Constantine Yurevich (CEO of SegmentStream) dropped a masterclass in media measurement.

One key idea: Most brands optimize for average ROAS, but what really matters is marginal ROAS.

Here’s the difference:

• Average ROAS = total return / total spend

• Marginal ROAS = return from your next dollar of spend

Why it matters:

• A campaign may look great at 4x ROAS on average

• But what if every additional $1 now only gets you $0.90 in return?

• You’re scaling into negative ROI without knowing it

📉 SegmentStream found that many brands do this, especially with channels like paid search and retargeting. The averages look healthy, but the diminishing returns are already killing profit.

Here are three signs your current attribution might be misleading you:

1. Your display or TikTok ads show 0 last-click conversions, but some model claims they drive 20% of sales

2. You “believe” in a platform’s contribution, but are afraid to shut it off and test it

3. You optimize for blended or average ROAS without any marginal testing

🎯 2 fixes you can apply:

1. Validate with testing:

• Scale spend (meaningfully) on a channel and see if overall revenue goes up

• Or cut spend (drastically) and see if revenue drops

• If nothing happens either way, it’s probably not incremental

2. Measure visit quality, not just conversions:

• SegmentStream uses behavioral modeling to track upper-funnel traffic quality

• Even if users don’t convert immediately, their engagement patterns can predict future purchases

Takeaway:

If you're diversifying away from Meta, remember: most platforms underperform in last-click. But that doesn’t mean they aren’t working. Model incrementality. Track marginal ROAS.

And don’t mistake noise for signal.

Watch the whole thing 👉 What DTC Brands Get Wrong About Attribution

🎯 TACTICS: Rebuilding the Creative Stack (With AI & Format Fluency)

Once you have your measurement dialed in, creative is your growth lever. But most brands are still stuck in one of two traps:

• Prompting AI like it’s a copywriting vending machine

• Making creative that screams “ad” in a world that only rewards “content”

This week, we’re breaking both habits—so you can ship faster, smarter, and more natively.

🤝 Use AI Like a Teammate, Not a Tool

In this podcast 👆 Stanford professor Jeremy Utley (co-author of Idea Flow) makes a provocative claim: “I don’t use AI. I work with it.”

In his research, teams that treated AI like a junior creative partner, not simply a productivity tool, produced 40% better output. And those who underperformed? They expected AI to do the work for them.

Here’s the shift:

• Low performers prompt once. High performers coach.

• Low performers ask questions. High performers let AI ask them better ones.

• Low performers take what they get. High performers refine, remix, and iterate.

💡 Tactical Drill:
Use this exact meta-prompt to flip the model:

🎧 Watch the full conversation here: Jeremy Utley on collaborating with AI

🧠 Creativity = More Than Your First Idea

One of the simplest insights from the podcast was also the most profound:

That’s the job: push past good enough. Because in a world where everyone’s using the same AI models, your inputs, iterations, and intent determine output quality.

Utley’s advice:

• Prompt for volume and variation

• Refine based on emotional tone, specificity, and edge

• Build systems to sort the best, not settle for first

🎙️ Native Creative That Doesn’t Feel Like an Ad

One consistent trend this year is make ads that don’t like ads on social.

Fraser Cottrell recently shared a high-performing podcast-style ad for Kolkata Chai. It wasn’t flashy. It wasn’t salesy. It just looked like two people talking about something they cared about.

Why it worked:

• Curiosity-first hook: “Did you know chai in India is completely different?”

• Education before the pitch: Product mentioned ~20 seconds in

• Format fluency: Podcast mics, casual cuts, UGC overlays, top/bottom splits

• Defined roles: Host vs guest for narrative clarity

🧵 Full breakdown: Fraser’s podcast ad thread. This breakdown and video will help you execute this style of ad for your brand.

🤖 Build That Format Using AI as Your Teammate

Here’s how to combine Fraser’s ad structure with Jeremy’s AI philosophy to produce creative that scales:

Step 1: Discover a Misconception

Step 2: Design the Back-and-Forth

Step 3: Explore the Range

Step 4: Visual Layering

Step 5: Refine Together
Coach it:

🧪 Layer Performance Analysis into the Creative Process

You can take this even further. Fraser recently shared another prompt on X that’s helping junior strategists operate like seasoned analysts:

Simple? Yes. But powerful.

By feeding the AI both the creative and performance metrics, it:

• Identifies overlooked patterns

• Connects creative moments to real conversion spikes

• Reinforces learning across your team

Fraser's full breakdown here.

🧠 Final Takeaway

AI is not another tool or shortcut. It’s a collaborator. And like any great teammate, the more you invest in working with it, the better your outputs become.

Treat creative like a system:

• Push for volume and variation

• Build with native formats in mind

• Train your model with context and feedback loops

• Layer in performance data for iterative improvement

Start with one curiosity hook. Build a dialogue. Coach your AI like a junior creative. Then test your hypotheses with real data.

Quick Hits:

• Notion introduces “Research Mode”, an AI that can act like your personal assistant.

• HBO rebrands “Max” to “HBO Max” and then laughed at itself about it on social.

• Anthropic releases Claude Research on mobile. 

• Did your Meta performance fall off a cliff this week? This might be why.