Discover why human judgment, creativity, and customer insight are becoming the strongest competitive advantages in AI-driven marketing.
For most of the last decade, advantage in marketing came from access. Who had the best data, the most advanced technology, or got to a new platform first. Generative AI has quietly ended that. The tools are now nearly universal, and they are largely the same tools.
The CMO Survey, run by Duke University's Fuqua School of Business with Deloitte and the American Marketing Association, put a number on how fast this happened. In January 2026, the 308 U.S. marketing leaders it surveyed reported that AI was used in 24.2% of their marketing activities, nearly double the 13.1% of 2024. Generative AI alone accounted for 22.4%, up from 7% two years earlier, and those same leaders expect AI to reach 55.9% of marketing activities within three years. The single most common use is content generation, cited by 73.9% of companies.
When everyone produces content with the same tools, the tools stop being an advantage. They become infrastructure, like spreadsheets or email. The question that matters now is how insight, creativity, and strategy are produced when every team starts from the same baseline.
When everyone has the same tools
When a tool goes universal, advantage moves.
| Tool | What it made easy | Where advantage moved |
|---|---|---|
| Spreadsheets | Calculation | Interpretation |
| Analytics | Measurement | Insight |
| Generative AI | Production | Judgment |
Five habits of teams that think with AI
- Use AI for research and synthesis, not as a content factory.
- Make your own knowledge usable, so AI reflects your context.
- Work in short learning cycles driven by strategy, not tools.
- Connect insight across search, content, paid and lifecycle.
- Protect judgment: design the human review step and train for it.
of companies now use AI for social listening and sentiment analysis, fall 2023 to 2026. The signals are easier to find. Deciding what they mean is still human work.
Source: The CMO Survey (Duke Fuqua, Deloitte, AMA), 2026: 14.2% in fall 2023 versus 28.2% in 2026.
Tool-based differentiation is ending, faster than before
Every widely adopted technology moves advantage from access to interpretation. AI is doing it in a few years.
This pattern has happened before. When spreadsheets became standard, financial advantage moved from doing the calculations to interpreting them. When analytics platforms became standard, advantage moved from collecting data to understanding it. Generative AI is following the same path, only much faster.
The productivity gains are real. Companies in The CMO Survey reported that AI in marketing improved sales productivity by 14.1% and customer satisfaction by 10.8%, and cut marketing overhead costs by 14.6%. Those gains explain why adoption has been so fast. They also explain the problem that came with it.
When production becomes cheap, volume stops signaling quality. AI did exactly what it promised: it made content faster and cheaper to produce. The side effect is that much of it now looks and sounds the same. Customers have noticed. In Gartner's September 2026 survey of 1,006 U.S. consumers, 65% said brands are producing too much AI-generated content. A team that publishes more of what everyone else is publishing will not stand out, however efficient it is.
Creativity moves upstream
AI is most valuable before the creative decision, not as a replacement for it.
A common worry is that AI will replace creativity in marketing. What is actually happening is more subtle. As execution gets cheaper, the most important creative work moves earlier in the process.
The strongest teams are not mainly using AI to produce finished assets. They use it to widen the range of questions they explore before deciding what to say. That includes summarizing customer research, finding themes across scattered inputs, testing different ways of framing a customer's problem, and checking how positioning language lands with different audiences. AI becomes a tool for exploration more than expression.
In that setting, creative advantage does not come from generating more ideas. AI can generate hundreds. It comes from recognizing which ideas are meaningful, credible, and consistent with what the brand stands for. That recognition does not come from the model. It comes from judgment.
Research becomes continuous, and small signals matter more
AI can surface early signals at scale. People still have to decide what they mean.
The most important change generative AI brings to marketing may be in research rather than execution. Traditional market research relies on large, structured inputs such as surveys, panels, and past performance data. Those inputs are still useful, but they arrive slowly. A quarterly brand tracker or an annual customer survey tells you what customers thought months ago, and by the time the report is presented, the market has often moved.
AI changes what is practical to study. Most of what customers actually say to a company is unstructured: sales call transcripts, support tickets, product reviews, search queries, community threads, and win-loss notes. Until recently, reading all of it was impossible for most teams, so it went unread. AI can now scan that material and group recurring themes in minutes. The CMO Survey shows teams moving in this direction: 28.2% of companies now use AI for social listening and sentiment analysis, double the 14.2% in fall 2023.
That makes research something a team can do continuously rather than in occasional projects. In practice, continuous research usually runs on a simple rhythm. Each week, AI tools group new customer inputs and flag what has changed. Each month, a strategist reviews those groupings and names what is genuinely new. Each quarter, leadership decides whether any of those signals should change positioning, messaging, or the product. That rhythm is a practical starting point, not a fixed rule, but it shows the shift: research becomes a steady input to decisions rather than a report that arrives after them.
The most valuable things this kind of research finds are small signals. A small signal is an early, easy-to-miss sign that something is changing. It might be a new word customers use to describe their problem, a new objection that starts appearing in sales calls, a competitor's name showing up in lost-deal notes, a search query that is suddenly growing, a point of friction that appears in one channel but not others, or a question buyers keep asking AI assistants about your category. None of these will move a dashboard on their own. Together, they often show where demand is heading before the numbers do.
The risk is treating every pattern as meaningful. AI will always find patterns, including ones that do not matter. Four questions help separate signal from noise. Does it recur over several weeks rather than appearing once? Does it show up in more than one source, such as both support tickets and reviews? Does it come from the customers you most want to win? And does it connect to a decision you could actually make? A signal that passes two or more of those tests deserves a person's attention.
This idea has a long history in strategy research. George Day and Paul Schoemaker's work on "peripheral vision" argued that companies that scan systematically for weak signals at the edges of their market tend to see major shifts earlier than those that watch only their core metrics. Lagging indicators, such as conversion rates, traffic, and campaign return on investment, confirm what has already happened. Small signals show what is starting to form. AI has made those signals far cheaper to collect, which changes the job of research. The hard part is no longer gathering information. It is deciding which of it deserves attention.
Judgment is now the scarce resource
When tools are shared, the ability to choose well becomes the advantage.
"Judgment" can sound abstract, so it helps to be specific about what it means in AI-assisted marketing. It comes down to four decisions that AI tools cannot make on a company's behalf. The first is what to say: which of many possible ideas fits the brand, the audience, and the moment. The second is what not to say: which claims the company can stand behind, which tones are off-brand, and which topics carry risk. The third is which signals matter: which of the patterns research surfaces deserve action. The fourth is when work is ready: when a draft is good enough to publish, and when it needs another pass or should not go out at all.
These decisions stay with people because they depend on context a model does not have. That includes the brand's history, the company's relationships with its customers, its legal and regulatory exposure, its long-term strategy, and, above all, accountability. Someone has to own the outcome when a message lands badly, and that cannot be delegated to a tool.
Consider a simple example. An AI tool drafts twenty headlines for a product launch in a few seconds. Three are genuinely strong. One makes a performance claim the legal team has not approved. Another sounds almost identical to a competitor's tagline. A third is technically accurate but would confuse the customers the launch is aimed at. Producing the twenty took no skill. Recognizing which three to use, and which three to stop, is where the value is, and it is the part of the work no competitor can copy by buying the same tool.
There is a gap here that most organizations have not closed. The same CMO Survey found that marketing training budgets have fallen to 3.8% of marketing spending, down from 5.8% before the pandemic, while AI-related capabilities, such as analytics, adopting AI into processes, and generative engine optimization, rank among the biggest barriers to getting value from marketing technology. Companies are spending more on tools while spending less on developing the people who direct them.
The organizations getting this right make judgment explicit rather than leaving it to instinct. They write down their editorial standards on a single page: voice, claims they will and will not make, and topics that need extra review. They clarify decision rights, so everyone knows who can approve what. They build a review step into the workflow at a set point, where AI-assisted work is checked against strategy before it goes out. And they develop judgment deliberately, pairing less experienced team members with senior reviewers and walking through rejected drafts together so the reasons become shared knowledge rather than one person's taste. None of this slows a team down. It keeps its speed pointed in a consistent direction.
The numbers behind the shift
AI in marketing, 2026
AI now supports 24.2% of marketing activities, up from 13.1% in 2024. Training gets 3.8% of marketing spend.
Marketing leaders expect AI in more than half of their activities within three years. The share of spend on training has fallen from 5.8% before the pandemic.
| Finding | Figure | Source |
|---|---|---|
| Share of marketing activities using AI | ||
| All AI, 2024 versus 2026 | 13.1% vs 24.2% | The CMO Survey, fielded January 2026 |
| Generative AI, 2024 versus 2026 | 7.0% vs 22.4% | The CMO Survey, 2026 |
| All AI, expected within three years | 55.9% | The CMO Survey, 2026 |
| How companies use AI in marketing | ||
| Companies using AI for content generation | 73.9% | The CMO Survey, 2026 |
| Companies using AI for social listening and sentiment analysis, fall 2023 versus 2026 | 14.2% vs 28.2% | The CMO Survey, 2026 |
| Investment in people | ||
| Training and development as a share of marketing spend, against 5.8% before the pandemic | 3.8% | The CMO Survey, 2026 |
| Consumer context | ||
| US consumers who say brands produce too much AI-generated content | 65% | Gartner, September 22, 2026 |
The CMO Survey, run by Duke University’s Fuqua School of Business with Deloitte, surveyed 308 US marketing leaders in January 2026. Activity shares measure how much of the marketing workload uses AI; company shares measure how many companies use AI for a task at all. The Gartner figure describes consumers, not marketers.
The CMO Survey reflects U.S. marketing leaders' own reports of activity and impact; it describes direction and scale, not audited results.
Turning the habits into workstreams
Habits fade unless someone owns them. A workstream gives each one an owner, a rhythm, and an output.
Most teams agree with the ideas above and then struggle to act on them, because "use AI for research" and "protect judgment" are habits, and habits without owners fade within a quarter. The practical step is to turn each habit into a workstream: a defined piece of ongoing work with a named owner, a regular cadence, and a clear output that someone else can use. Here are five, with a first step for each.
1. Signal research. This workstream turns raw customer input into a short list of what is changing. The owner is usually an insights or strategy lead. AI tools group new inputs weekly, and a person reviews them monthly. The output is a one-page "what's new" brief that names emerging signals and says which ones deserve action. The first step is connecting the sources: call notes, support tickets, and reviews in one place the AI tool can read.
2. Company knowledge. This workstream makes your own experience usable, so AI tools draw on your context rather than generic knowledge. The owner is often a marketing operations or content lead. The cadence is monthly additions. The output is an organized library of past research, campaign results, customer feedback, and key strategic decisions. The first step is gathering the five documents your strategists refer to most and putting them in one well-labeled place.
3. Learning sprints. This workstream answers one specific question at a time. Ownership can rotate, with a different team member leading each sprint. A sprint of two to four weeks is a practical length. The output is an answer and a decision, such as which message to lead with for a new segment, or where buyers drop off and why. The first step is choosing one question from the latest signal brief.
4. Cross-channel insight. This workstream makes sure what one channel learns reaches the others. The owner is the head of marketing or growth. The cadence is a short monthly meeting with the search, content, paid media, and customer lifecycle leads. The output is a shared list of learnings that each channel commits to act on. We cover how this works in more depth in "A Guide to Connected Marketing."
5. Standards and review. This workstream protects judgment as a capability. The owner is a brand or editorial lead. The review step applies to every AI-assisted asset, and the standards themselves are updated each quarter. The outputs are the one-page standards, a simple log of what was changed or rejected and why, and regular sessions where the team reviews those decisions together. The first step is writing the one-page standards.
The AI-era marketing operating model
Five workstreams turn AI from a tool into a habit. Each needs an owner, a rhythm and something to show for it.
| Workstream | Typical owner | Cadence | Output |
|---|---|---|---|
| Signal research | Insights or strategy lead | Weekly grouping, monthly review | A one-page “what’s new” brief. |
| Company knowledge | Marketing operations or content lead | Monthly additions | An organized library AI tools can draw on. |
| Learning sprints | Rotating sprint lead | Two to four weeks per sprint | An answer and a decision. |
| Cross-channel insight | Head of marketing or growth | Monthly | Shared learnings each channel acts on. |
| Standards and review | Brand or editorial lead | Every asset; standards updated quarterly | Standards, a review log and training sessions. |
Owners are typical roles, not required hires. In a small team, one person can cover several workstreams.
A small team does not need five people for this. One person can own two or three workstreams, and most teams should start with signal research and standards and review, since those two most directly support the others. What matters is that each workstream has a name next to it. Unowned work is the first thing to disappear when a busy quarter arrives.
What this looks like in practice
Five habits separate teams that think with AI from teams that just produce with it.
Treat AI as a research and synthesis layer, not a content factory. Use it to map markets, group customer insights, and surface small signals that inform positioning and creative direction.
Make your own knowledge usable. Organize past research, campaign results, customer feedback, and strategic decisions so AI tools can draw on your company's context rather than generic knowledge. A team whose AI knows its history will produce work no competitor can copy.
Work in short learning cycles, not tool experiments. Design focused sprints to answer specific questions about customer needs, message clarity, or friction in the journey. AI speeds these cycles up, but the questions should come from strategy, not from whatever the newest tool can do.
Connect insight across channels. Let what you learn in search, content, paid media, and customer lifecycle data inform each other, rather than optimizing each channel on its own. We cover how this works in "A Guide to Connected Marketing."
Protect judgment as a capability. Make human review a designed step, not an afterthought, and invest in training the people who perform it. That is what keeps value compounding instead of dissolving into volume.
What this likely means over the next 6 to 18 months
Expect AI usage to keep rising, and the gap between teams to widen.
If marketing leaders' own projections hold, AI will handle more than half of marketing activities within three years. That would make AI-produced content even more common, and it would likely make undifferentiated content even less effective. The teams most likely to pull ahead are the ones that invest in research, interpretation, and clear standards alongside their tools, because those are the parts of the work that cannot simply be bought. As AI search tools increasingly summarize what is said about each company, a distinct point of view grounded in real customer insight is also likely to become more visible than generic content, not less.
If you want help turning your customer signals into clearer positioning, Camino5's Consumer Journey Strategy work starts there.
How to start today
List the three AI tools your team uses most, then ask whether any competitor could use the same tools in the same way. If the answer is yes, those tools are infrastructure, not advantage.
Pull ten recent sales call notes or support conversations and ask an AI tool to group the recurring themes. Then have a person on your team pick the one theme that is new this quarter and decide what it means.
Write down your editorial standards on one page, including tone, claims you will and will not make, and who approves what. Use it as the review step for all AI-assisted content.
Gather your best past research and campaign learnings into one organized place your team's AI tools can draw on. Start with the five documents your strategists reference most.
Compare what your team spends on AI tools this year with what it spends on training the people who use them. If the ratio surprises you, raise it in your next budget discussion.
The organizations that succeed in this environment will not be the ones that use AI the most; they will be the ones that think most clearly with it.
Sources
George S. Day and Paul J. H. Schoemaker, Peripheral Vision: Detecting the Weak Signals That Will Make or Break Your Company, Harvard Business School Press, 2006