Execution Is Now Cheap. Outcomes Are Now Scarce. Nobody Told Your Brand.
70% of companies are deploying AI in marketing. If everyone has the same tools, running faster isn't a strategy. It's a treadmill.
Since Issue 4 I’ve been promising to write the brand moat article.
The reason I kept delaying: I wanted to wait until the data was definitive rather than directional. It’s definitive now.
The 2026 CMO Survey - conducted in early April among senior marketing leaders - landed this week.
Headline: marketers are at their highest level of pessimism since 2020.
Budgets shrinking. Scrutiny rising. AI adoption accelerating. And here’s the sentence that stopped me: despite 70% of companies actively deploying AI in at least one marketing function, the vast majority report no meaningful impact on business outcomes.
More tools. Less confidence. No clearer answer to the question that matters: what actually differentiates a brand now?
I’ve been thinking about that question for 30 years. I finally have an answer I’m willing to put in writing.
What just happened to differentiation
For most of the last two decades, marketing differentiation lived in three places.
Speed. The brand that published faster, responded faster, iterated faster. The one that got the campaign out while competitors were still in approval.
Scale. The brand that could produce more content, reach more channels, run more variants, test more hypotheses simultaneously.
Craft. The brand with better creative - sharper copy, stronger visuals, more coherent narrative, more distinctive voice.
AI just commoditised all three. Simultaneously.
Speed: any competent marketing team with an AI stack can now produce in hours what used to take weeks. Scale: AI can generate hundreds of variants, run them across every channel, and optimise in real time without a media buyer touching a keyboard. Craft: the average output quality of AI-generated marketing has crossed the threshold of “good enough” in most categories. Not exceptional. Good enough. Which in a world of abundant content, is often all that’s required to compete.
The CMO Survey confirms what practitioners already feel: when everyone has the same acceleration tools, acceleration stops being an advantage. You’re all moving faster. The relative positions haven’t changed. You’ve just increased the speed of the treadmill.
This is the brand moat crisis. And most marketing leaders are responding to it by asking for more AI tools.
The Goldman number that reframes everything
Goldman Sachs published research this quarter that created a brief panic in boardrooms before being quietly filed and forgotten.
They surveyed nearly 6,000 executives. They found that 70% of companies are actively using AI. They also found that approximately 80% of those companies report no meaningful impact on employment or productivity. Zero. Not marginal. Not incremental. Zero.
But here’s the nuance that got lost: Goldman did find significant productivity gains in exactly two places. Software development and customer service - both showing a median improvement of around 30%. Everywhere else: no meaningful relationship between AI adoption and output at the economy-wide level.
The researchers at the National Bureau of Economic Research called it a “productivity paradox” - perceived gains are larger than measured gains. Companies feel more productive. The numbers don’t confirm it.
The reason, as Goldman’s analysts explained: the gains are real but trapped. AI makes individual workers faster. That speed doesn’t automatically make the surrounding system - the supply chain, the pricing model, the customer experience, the revenue line - more efficient. The productivity is happening inside the team. It isn’t escaping into the business.
For CMOs and brand leaders, this is the most important finding of the year. And it’s being ignored.
Your AI tools are almost certainly making your team faster. The question is whether that speed is escaping into something customers can feel. If it isn’t - if your brand is producing more content faster but the content is less distinctive, less specific, less human - then you’ve optimised for output while your actual competitive position has deteriorated.
Faster noise is still noise.
What Pernod Ricard found that should worry you
Harvard Business Review published a piece this March on preparing brands for agentic AI. One detail inside it is more alarming than anything in the headline.
The head of digital at Pernod Ricard - a serious, sophisticated brand operator - commissioned a study on what AI models were saying about his liquor brands. The finding: LLM data was often incomplete or incorrect. One popular AI model miscategorised Ballantine’s Scotch whiskey - an affordable mass-market offering - as a prestige product.
Think about what that means. An AI agent helping a consumer choose a whisky gift, or a buyer sourcing stock for a hotel bar, would recommend Ballantine’s in the wrong context, at the wrong price point, to the wrong customer. The brand positioning that Pernod Ricard has spent decades and hundreds of millions building - the specific, calibrated place Ballantine’s occupies in the market - simply doesn’t exist inside the model’s representation of the category.
Two thirds of Gen Z and more than half of Millennials now use LLMs to research products before buying. Your brand positioning, as understood by the AI systems that are increasingly mediating those decisions, may bear no relationship to the positioning your marketing team believes it has established.
This is a new category of brand risk. And it doesn’t show up in any traditional brand tracking metric.
The three things that are actually scarce now
If speed, scale, and average craft are commoditised - and the data says they are - then differentiation has to come from somewhere else. Here’s what’s actually scarce in 2026.
Specificity. The single hardest thing for AI to generate is genuine specificity - the particular detail, the precise observation, the exact truth about a specific customer in a specific situation. AI generates fluent generality. The brand that makes specific choices - this customer, this moment, this truth, expressed this way and no other way - is producing something that can’t be replicated by anyone running the same tools with a different brief.
Specificity isn’t just a creative principle. It’s a strategic one. The brand that stands for something specific is also the brand that AI models learn to represent accurately, because there’s something coherent to represent. Vague brands - the ones that stand for “quality” and “innovation” and “customer-centricity” - are the ones that AI miscategorises, because the model has nothing distinctive to hold onto.
Courage. Every AI optimisation system, by design, moves toward the centre. It tests variants, identifies what performs, amplifies what works, eliminates what underperforms. Applied consistently across an industry, this produces convergence - brands that look increasingly similar because they’ve all been optimised toward the same performance signals from the same customer behaviour data.
The brand that makes choices that don’t optimise - that sacrifices short-term performance for long-term distinctiveness, that says something true rather than something that tests well, is increasingly rare. Rare is differentiated. Courage is a moat in an industry of optimisation machines.
Presence. There is a difference between content and presence. Content is what you produce. Presence is what people feel when they encounter your brand, the sense that there’s a coherent, intentional, specific point of view behind what they’re seeing. Presence can’t be automated because it has to be consistent across every touchpoint, over time, including the ones that don’t have a brief attached. It’s built in the accumulation of specific choices, not in any single piece of content.
The 2026 CMO Survey found that AI adoption accelerates content production while marketing pessimism rises simultaneously. That’s not a paradox. That’s what happens when you produce more content while presence erodes. More output, less brand.
The agent visibility problem and why it’s urgent
There’s a new dimension to brand building that didn’t exist 18 months ago and most brand teams aren’t addressing.
Your brand now needs to be legible to machines, not just humans.
When an AI agent is helping someone choose a supplier, book a service, or evaluate a product category, it’s not seeing your advertising. It’s not experiencing your brand narrative. It’s querying structured information, your reputation in the training data, your representation in review platforms, your clarity on pricing and positioning and credibility signals. The agent makes a decision about whether to include you based on how well it can understand what you are, what you do, and whether you’re trustworthy, in machine-readable terms.
The HBR piece calls this “share of model” - the percentage of AI-mediated category queries in which your brand is accurately represented and recommended. You don’t have a metric for this yet. Your agency doesn’t track it. Your brand health tracker doesn’t include it.
But your competitors are starting to think about it. And the brands that establish clear, coherent, machine-legible identities now - while the category representations in AI models are still forming - will be far harder to displace than the brands that try to correct misrepresentation after it’s been trained into the system.
You cannot buy your way into a model’s representation of your category. You can only earn it, through the clarity and consistency of what you are.
What this actually means for your next decision
Stop for a moment. Not to plan. To audit.
Take the last ten pieces of marketing your team produced. Put them side by side. Ask one question: if I removed the logo, could someone identify this as ours?
If the answer is mostly no - if the content is competent but interchangeable, fluent but generic, produced at speed but lacking the specific signature of your particular point of view — then you have a brand moat problem. And no amount of AI tooling will fix it. In fact, more AI tooling will accelerate it.
The tools are not the problem. The absence of something specific to express is the problem.
Your next brief shouldn’t ask: how do we produce more, faster? It should ask: what is the one true thing about our brand that only we can say? And is every piece of content we produce an expression of that thing, or a distraction from it?
In a world where execution is cheap, the only scarce thing is a reason to choose you specifically.
Find that. Protect it. Say it with enough conviction that even a machine can’t get it wrong.
Next: The CMO is now the most dangerous job in business. Here’s what the data actually says - and what 30 years on the producing end of this tells me about what comes next.
Further Reading
CMO Survey 2026, Deloitte / Duke Fuqua / AMA (April 2026);
Goldman Sachs / NBER, AI Productivity Paradox research (March 2026);
Harvard Business Review, Preparing Your Brand for Agentic AI, Acar & Schweidel (March–April 2026);
MarTech, How AI Agents Will Reshape Every Part of Marketing in 2026 (January 2026); Adweek, 10 AI Marketing Trends for 2026 (February 2026).
Undercurrent is a weekly newsletter for business leaders navigating the agent economy. No jargon. No hype. Just the shifts that matter, explained clearly.
If this made you think of someone who needs to read it: Forward it. That’s how this grows.
Written by Alemsah — Chief Storyteller, Disrupt.