AI & Marketing Technology

Three Campaigns That Couldn't Have Been Briefed Two Years Ago

Three specific campaigns reveal the real gap in AI adoption: not what's technically possible, but what different organizations can even conceive of. And the frontier's already moved on.

There's a difference between making things faster and making things possible that didn't exist before. Most AI stories miss this distinction entirely. They're about execution—cutting weeks into days, automating what humans used to do manually. Good for efficiency, fine for ROI conversations. But the campaigns I want to talk about aren't about speed. They're about imagination.

The real story is this: there are campaigns running right now—specific, winning, generating real revenue—that a strategic planner in 2023 couldn't have briefed if they wanted to. Not because the technology didn't exist. But because the organizational infrastructure to conceive of them didn't exist. And that gap is widening faster than most CMOs realize.

Let's start with the campaign everyone's citing but no one's actually examined closely: Coca-Cola's 2025 holiday work. The headlines read like a Silicon Valley fever dream: "70,000 AI clips in 30 days with five specialists, replacing crews of 50+ over 12-month cycles." I've seen that stat everywhere. But here's what nobody asks: what does "70,000 AI clips" actually mean?

This matters because the methodology isn't transparently documented. Are these 70,000 entirely unique, fully rendered scenes? Unlikely. What Coca-Cola actually did—and this is the crucial distinction—was create a production system that generates variations on template-based creative environments. The available reporting from The Wrap suggests they've built a system where foundational scenes—a kitchen, a living room, a snow-covered landscape—can be customized at speed. Within those environments, you layer different product placements, different lighting conditions, different seasonal contexts, different cultural references. It's template-based assembly at massive scale, not 70,000 independent full-CGI renders from scratch.

I'm flagging this explicitly because the ambiguity matters for how you think about this work. Are they generating 70,000 completely novel environments? Are they assembling contextual variations within established templates? It's probably somewhere between. But that distinction—hybrid approach vs. pure generative capability—changes how you think about what's replicable in your own organization.

Why does this distinction matter at all? Because it reframes the actual conceptual innovation. In 2023, the briefing would have been impossible for a different reason: you couldn't promise a creative team it was feasible to test that many environmental variations without driving costs into the stratosphere. You'd have needed a production crew that stayed stable for twelve months, multiple environment builds, lighting rigs, all locked in before you knew which combinations would actually resonate. The innovation here isn't "AI generates everything automatically." It's "we can now explore the entire permutation space of how our product sits in these contexts"—and that's a fundamentally different strategic question than "how do we produce 50 ad variations from a single hero shoot?"

The briefable part in 2023 was: "Make us one killer hero spot that works everywhere. Test it in market. Iterate if needed." The unbriefable part was: "Generate 70,000 contextual variations systematically, cluster them by performance, and let us see which environments, lighting conditions, and cultural contexts actually move the needle for our audience." That's the imagination gap. One is execution velocity. The other is strategic scope.

Corona's Pinterest campaign is quieter but more conceptually audacious. The campaign generates photorealistic contextual backgrounds for Corona products—beach scenes, backyard setups, resort contexts, tropical vacation moments—and assembles them into platform-optimized video content in real time. Not once. Not 70,000 times. Continuously.

Think about what you'd have to brief a creative director in 2022. "We're going to generate custom summer and vacation contexts for every variation of our product, in real time, based on trending topics and seasonal moments within Pinterest, and deliver them to users in the format they're actually watching content in on that platform." You'd get blank stares. Not because the idea is bad—it's genius. But because the operational requirement would have been to build a production pipeline with this kind of capability, and you couldn't articulate what that would even look like before 2023.

Here's what infrastructure Corona actually needed to make this work:

First, generative scene creation that maintains brand consistency while generating variations. You're not creating random beach scenes. You're creating scenes that feel like Corona contexts—warm, inviting, aspirational—while being customized for different seasonal moments and cultural contexts. That requires a generative model trained on Corona's specific visual language.

Second, real-time asset assembly that doesn't require human review for every output. In 2023, this would have meant submitting every generated scene to a brand team for approval before publishing. That kills the real-time part entirely. You need automated quality gates that catch brand-misaligned or problematic outputs without human intervention per asset.

Third, platform-specific rendering that optimizes for how people actually consume video on that channel. Pinterest video has different aspect ratios, timing preferences, and engagement drivers than TikTok or Instagram. You need to render the same contextual scene into different formats for different channels without rebuilding the entire asset each time.

In 2023, that's a six-month build with technical debt, vendor lock-in risk, and the fundamental problem that nobody knows if the system will actually work at production scale. You'd be writing a scope document that's more speculative fiction than product brief. Today, this is buildable. But only if you have the organizational readiness to invest in it.

This is the shift: Corona didn't just execute a campaign faster. They conceptualized something that required AI-native creative infrastructure to even imagine. The brief changes from "create ads that hit these seasonal moments" to "build a generative engine that produces contextually relevant assets for a continuous audience." That's not speed. That's a completely different strategic layer.

Burger King's "Million Dollar Whopper" campaign takes a step further into economic inversion. Fans design custom Whopper variations. Then—this is the part most people miss—AI generates a custom jingle for each one. Not a template. Not a variation on an existing sonic brand element. A unique, custom audio asset that still lands within the Burger King brand sonic universe and reinforces their positioning.

Let's think about briefing this in 2022. "We're going to have customers co-create products, and then we're going to generate unique audio signatures for each custom variation that reinforces brand positioning and emotional resonance. And we're going to do this at scale for hundreds or thousands of customer designs." Okay, sounds cool. Now, here's the reality check: how much is that going to cost per jingle? In 2022, you're looking at:

Months of work with composer(s). Iterations and revisions for each custom variation. Brand reviews and approval cycles for every audio asset. Rights clearance and production logistics. You'd be looking at $500-$5,000 per custom jingle at minimum, depending on production quality. For 1,000 customer designs, that's $500,000-$5,000,000 in production budget. For a campaign that might get viewed by 5,000-50,000 people in total.

The budget doesn't work. The ROI doesn't justify it. The entire campaign concept becomes economically irrational in 2022. That's not a creative problem. That's a unit economics problem that makes the entire idea unbriefable.

What changed with AI isn't that we can make jingles faster. It's that we can now afford to create a completely custom sonic asset for something that might only get viewed by 5,000 people instead of 50,000,000. The cost per asset dropped by an order of magnitude. The economics of scale inverted. And that inversion—that fundamental change in what's economically reasonable to customize—opens up entirely new campaign architectures that didn't exist when every asset had to earn its economic weight through massive reach.

Burger King created a campaign concept that operates on a completely different unit economics model than anything they could have briefed pre-2023. That's not execution innovation. That's conceptual innovation enabled by economic transformation.

The real innovation is economic. When the cost per custom asset drops by an order of magnitude, campaign concepts that were economically irrational become strategically possible.

Here's the number that should keep CMOs up at night: only 22% of organizations have established AI governance guardrails. That means 78% of companies are still operating in exploratory mode, ad-hoc implementation, pilot projects without organizational infrastructure. And that structural difference—not in technology, but in organizational readiness—directly determines what campaigns can even be conceptualized.

This isn't abstract theorizing. Let me make it specific. To brief the Corona campaign, you need a marketing organization where strategy is completely decoupled from execution. A strategist owns the narrative direction and guardrails. Then autonomous systems generate variations within those rails. That requires trust in AI outputs that most organizations don't have yet—and governance structures to oversee it without creating bottlenecks.

You also need data maturity that enables real-time performance feedback. If you're going to generate assets continuously and assemble them based on what's actually landing with the audience, you need to feed performance data back into the creative system within hours, not weeks. Most organizations don't have that data pipeline. The data lives in disparate systems. Reporting lags by days. By the time you know what worked, the campaign moment has passed.

You need creative leadership that's been fundamentally reimagined. Instead of hiring people who execute a director's vision—people with strong technical skills in motion, color grading, composition—you're hiring people who can define strategy at a higher level of abstraction and oversee AI-driven variation generation. That's a different skill set entirely. Most creative leaders came up through execution, not abstract system design. Their entire career has been about hands-on craft. Now you're asking them to become creative infrastructure designers instead.

You need cultural permission to experiment with autonomous systems. In conservative organizations—and most organizations are conservative about their brand—the idea of letting AI generate content without human review of every output triggers legitimate governance fears. That's not irrational. It's reasonable caution. But it also makes certain campaign categories fundamentally unbriefable in that organization.

When Coca-Cola and Corona and Burger King started building these campaigns, they had organizational maturity in these areas. The org structure supported it. The approval processes could handle scale. The data systems fed back into creative in real time. When the median organization—the 78% without governance structures—tries to build similar work, they hit structural constraints that aren't technical. They're organizational, cultural, systemic.

Here's where it gets uncomfortable. Deloitte's research on AI adoption in the enterprise—specifically from the State of AI in the Enterprise 2026 report—shows that frontier firms and median organizations are diverging at an accelerating rate. Frontier firms send approximately 2x the volume of AI-driven messages and operate with 6x the depth of integration compared to median firms. That's not small. That's a structural capability gap.

I want to be careful about how I frame this—it's not that frontier firms do 2x more volume because they have bigger budgets or more headcount (though that's part of it). It's that deeper organizational integration creates feedback loops and compounding advantage. Marketing learns from finance's AI implementation. Finance learns from operations. Sales learns what's working in marketing and applies principles to their pipeline. Best practices compound across functions. Meanwhile, median organizations are piloting AI in functional silos. Marketing tries a gen-AI tool for copy. IT runs a separate pilot for process automation. Finance explores AI for forecasting. They never talk. Best practices don't compound. What you get is slower progress—and qualitatively different progress.

What does that mean for campaign imagination? It means the kind of campaigns you can conceive. A Coca-Cola campaign relies on real-time performance feedback loops between creative generation and media performance. If your organization hasn't integrated AI into finance (to understand unit economics and ROI at scale), operations (to manage creative workflows and asset approvals), and marketing (to measure creative effectiveness and feed it back to generation), you can't even brief a campaign that depends on that integration. The concept becomes impossible without the infrastructure.

The frontier is getting further away faster. Not because of technology breakthroughs. Because of organizational adoption depth. Frontier firms are compounding advantage across multiple functions. Median firms are still figuring out what AI is. And that means the campaigns you can brief change based on which adoption stage your organization is in.

The frontier is getting further away faster—not because of technology, but because organizational adoption depth creates compounding advantage.

If 73% of marketing teams now use generative AI in their workflows, that should mean campaign imagination is normalized everywhere. The adoption should be broad. But here's the caveat that fundamentally changes how you read that stat: the 73% figure from Loop Ex Digital reflects any use of generative AI tools in a given week. The survey doesn't distinguish between exploratory experimentation and systemic adoption. That includes prompting ChatGPT once to brainstorm email subject lines. That includes running copy through an AI writing assistant one time. That includes asking Midjourney to generate a hero image for a presentation.

That's critical context. Because the question isn't "do you use AI?" It's "what organizational infrastructure do you have to use AI in campaign conception and execution at scale?" And that's maybe 15-20% of the 73%. The gap between "I've played with generative AI tools" and "my organization has fundamentally reimagined what campaigns we can brief and execute" is cavernous.

There's a counterweight to all of this frontier-firm optimism. And I need to name it because it's real. 59% of consumers distrust AI-generated content unless transparently disclosed. That's not a small margin. That's a majority-trust-deficit for anything that looks like fully AI-generated assets.

What does that mean for the campaigns I've been describing? The successful ones—Coca-Cola, Corona, Burger King—are managing this by having humans in the loop. Coca-Cola's clips have human oversight and brand-team approval built in. They're positioned as "variations on this creative concept," not "AI generated this autonomously." Corona's contextual assets are supplemented with human-directed content. Burger King's AI-generated jingles are reviewed and approved. They're not trying to hide the AI—they're creating a hybrid architecture where AI handles scale and variations while humans own strategy, approval, and brand voice.

The briefable campaign isn't "let AI do everything autonomously." It's "use AI as infrastructure while humans own strategy, direction, and approval." Which is actually another layer of organizational constraint. You need humans who understand how to set creative guardrails and oversee outputs at scale without becoming a bottleneck. That's a different skill set than traditional creative direction. It's closer to art direction through AI systems than hands-on execution. And it's rarer.

The campaigns that are working aren't fully autonomous. They're human-directed systems powered by AI infrastructure. That distinction matters because it means the organizational skill you actually need is strategic direction at scale—the ability to define constraints and vision in a way that an AI system can operationalize and vary. That's not the same skill as executing a creative director's vision. It's not the same skill as running a production. It's higher-level strategy applied to creative systems. And that's even rarer than pure creative talent.

I know what you're thinking: "I'm not Coca-Cola. I don't have that infrastructure or that scale." Fair. So what can you actually brief today that was impossible in 2023? What campaigns exist at your adoption stage?

You can brief campaign concepts that involve real-time optimization of creative elements based on live performance data. The infrastructure is simpler than Coca-Cola's. Dynamic Creative Optimization now assembles personalized variations within milliseconds for individual users, allowing you to test different headlines, images, body copy combinations, and value propositions on a per-user basis. Not 70,000 static variations. Live, responsive personalization that adapts based on user signals and behavioral data.

You can brief campaign concepts that use AI to simulate audience response at ideation stage. Instead of showing 20 concept variations to a 12-person focus group and praying they're representative, you can use customer digital twins to validate ideas at scale before execution. That collapses the time between "crazy idea" and "data-backed decision" from weeks to days. You can test a campaign concept with a simulated audience of 10,000 personas before you spend a dollar on production.

You can brief campaign concepts that use autonomous agents to handle multi-step campaign planning and optimization. Instead of a campaign planner spending weeks sequencing when different channels launch, which creative variations go to which audience segments, and how to optimize based on early performance—you describe the strategic intent and the agent figures out execution. That compresses a month of planning into a week of oversight and human decision-making on strategic inflection points.

What all of these have in common: they're possible today if your organization has the governance infrastructure and data maturity to support them. They're not possible if you're still in the "let's play with ChatGPT and see what happens" stage. And that's the real dividing line between what you can brief and what remains conceptually out of reach.

The Campaign Briefing Spectrum: What You Can Conceive Based on Adoption Stage

Exploratory Stage: AI as a tool for faster brainstorming and copy generation. Campaign concepts unchanged from 2023. Examples: AI-assisted subject lines, ChatGPT ideation sessions.

Systemic Adoption Stage: Real-time creative optimization and audience simulation. Campaign concepts shift toward data-driven personalization. Examples: Dynamic Creative Optimization, digital twin validation, performance-driven asset selection.

Frontier Integration: Autonomous planning, continuous variation generation, human-directed AI infrastructure. Campaign concepts fundamentally reimagined around scale without compromising personalization. Examples: Coca-Cola's systematic variation generation, Corona's real-time contextual rendering, Burger King's custom audio generation.

I started this by saying the real story isn't speed. It's imagination. But I was imprecise. The real story is this: your organization's imagination is constrained by your adoption infrastructure, not by technology or creativity.

In 2023, a CMO at Coca-Cola and a CMO at a mid-sized brand would have briefed campaigns that looked identical on the strategic level—reach the right person with the right message at the right time. The execution was different, but the conceptual framework was the same. Today, that frontier CMO is briefing something that didn't exist as a category two years ago. Systematic variation generation. Real-time contextual assembly. Dynamic unit economics that invert scale requirements.

Meanwhile, the mid-sized brand is still briefing campaigns in 2025 that would have been conceptually identical to 2022. The tools have changed. The process speed has improved. But the campaign category—the fundamental strategic architecture—is the same. Why? Because the organizational infrastructure isn't there.

The technology is available to both. The frontier firm and the median firm can both license the same AI tools. But organizational readiness isn't available equally. And readiness determines imagination. Readiness determines what you can conceive of. Readiness determines what you can brief.

That's the real divide. It's not technical frontier vs. laggards. It's organizations that have fundamentally restructured how they work with AI—how they think about creative strategy, how they organize approvals, how they integrate data, how they deploy creative teams—versus organizations that are bolting AI onto existing processes. One is opening new campaign possibilities. The other is making existing campaigns slightly cheaper and faster.

The CMOs who win in 2025 and beyond won't be the ones with access to the best AI tools. They'll be the ones who've restructured their teams, their governance, their data flows, their approval processes, and their creative vision work around what AI infrastructure makes possible. That's organizational transformation, not technological adoption. And it's already happening. If you're waiting for better tools before you start that work, you're already behind.