In 1987, Nobel prize-winning economist Robert Solow wrote, “You can see the computer age everywhere … except in the productivity stats.” The maxim became known as Solow’s paradox: an increase in computer usage didn’t necessarily translate into an increase in good work.

 

The same principle applies to AI and marketing. In 2025, the B2B marketing world once again experienced the stark difference between output and outcomes.

 

Across B2B technology companies, AI adoption surged past 80%, but fewer than 1 in 5 teams could tie those tools to clear, measurable business outcomes. Cycle times shortened by as much as 40%, while the signal-to-noise ratio inside the content pipeline dropped just as fast, meaning that marketers were producing more static and yielding less engagement faster than ever.

 

Output and efficiency may be increasing. Outcomes and effectiveness … not so much.

 

For many teams, the story is the same. Dashboards are glowing. Content is multiplying. Performance reports are stacked higher than ever. But the more the machines produce, the more the brand’s distinction fades. What once sounded unmistakably human begins to feel algorithmic. Customers don't revolt—they simply disengage.

 

That is the quiet consequence of automation without discernment: output rises while resonance falls. The first phase of AI in marketing was defined by possibility—looking at what we could automate. The next phase will be defined by selectivity—figuring out what we shouldn’t automate.

 

This has enormous implications on marketing budgets and planning. For B2B tech companies, marketing budgets range anywhere from 7% to 12% of revenue, depending on which research you look at it. Overall, however, it does seem to be up slightly from the year before—roughly a 7% increase, by some accounts.

 

But if you look a little closer at those numbers, you’ll see a subtle shift. Investment in AI tools has grown to nearly 20% of total marketing spend, while the share devoted to creative and strategy has plateaued. Leaders seem to sense the imbalance. 81% of B2B marketers use AI in daily work, but most admit it has not yet delivered proportionate value. The plateau isn’t technical; it’s human. Scaling automation without making sure it’s squarely centered around human judgment leads to diminishing returns.

 

This imbalance shows up most clearly during budget season. CFOs want attribution and proof of ROI when it comes to marketing budgets. That can be tricky for marketing leaders, who are left trying to explain the much-harder-to-quantify value of influence, reputation, customer engagement and trust. On the other hand, thanks to a flood of hype and over-promise around AI, hard stats abound around potential efficiency gains from AI.

 

As famed sociology professor William Bruce Cameron once wrote, “Not everything that can be counted counts, and not everything that counts can be counted.” (The quote has often been attributed to Albert Einstein, but he was just quoting Cameron.)

 

The point is, just because the finance department sees hard numbers for AI investments and often “squishy” data points around the impact of human marketing work, that doesn’t mean AI is always the smarter bet. Quite often, the quantitative promise of AI in marketing falls short. So how do you account for that in the ROI calculations?

 

The most successful B2B tech marketing leaders aren’t the ones promising certainty or trying to concoct an artificially boosted ROI on marketing investment. Nor are they ones betting the farm on AI while reducing marketing headcount.

 

The most successful B2B tech marketing leaders are the ones who find the truest thing to say about marketing in the age of AI, which is “We’re figuring it out.”

 

They build their cases as structured experiments: start with pilot investments, measure specific gains, expand only when evidence warrants. A phased approach is not a hedge; it’s a discipline. It signals accountability, maturity, and an understanding that progress must be earned before it’s scaled.

 

This same discipline applies to the human-in-the-loop marketing in an AI-dominated world. Teams that treat AI as an advisor, not an operator, consistently outperform those who either avoid AI altogether or delegate too much to AI. One global software company that combined AI analytics with human narrative design saw a 30% lift in engagement. Another recorded a 25% increase in repeat interactions after reinstating human review into automated campaigns.

​

These outcomes are not anomalies; they’re indicators of a deeper truth. Those who recognize that marketing—the act of emotionally and commercially connecting with customers—is a distinctly human domain, will have an advantage over others who are too quick to dehumanize and over-automate it.

​

I also want to emphasize, we at Rob Roy are not anti-AI. Quite the opposite! Over the past year, we’ve been working with a custom GPT that we built within OpenAI. It helps us with everything from messaging analysis and content creation to bespoke sales talk tracks and competitive battlecards. However, our GPT is grounded in human-centric frameworks for persuasion, storytelling and authenticity, and at no point in time are humans excluded from the process. It’s a tool we use selectively and strategically, and it’s one that still relies on us for an intuitive understanding of what we’re trying to accomplish, and why.

 

So this budgeting season, the question for every B2B marketing leader is where and how to anchor the human in the loop of your AI-enhanced marketing processes. Start by asking yourself where things like human empathy, intuition, creativity and discernment are most needed. Look at where you’re experiencing the most friction in your marketing funnel, and start there. Look at where your people are running up against time constraints, and determine where automation could free them up to focus on things an AI would be less effective at.

 

Then build in guardrails. Determine which decisions and outputs require human validation before they reach the market. Determine what you do NOT want to entrust to an AI. These guardrails won’t slow you down. Often times, they’ll save you the wasted time and expense of recovering from failure later.

 

When your guardrails don’t prevent failure, build in mechanisms for rapid feedback, candid and courageous diagnosis of what is and isn’t working, iterate and move on. Look, somewhere between 70% and 90% of all AI deployments fail. Chances are, you’re NOT going to get it perfect the first time. Or the second time. Build experimentation and learning into your budgeting logic. Start small. Define success criteria. Learn faster than your peers. Have honest conversations with your finance teams around which experiments are paying off, which aren’t, and what you’re learning along the way that will pay off later.

 

In other words—budget for learning. Accelerating your learning curve is mission-critical in the era of AI, and that education isn’t free. Budget for it.

 

By next year, marketing budgets will likely increase again, with AI commanding a steady share of that increase. The companies who thrive will be the ones who were honest about the experimental nature of AI in marketing. Machines will continue to produce at scale, but people will continue to determine meaning. The competitive edge will belong to those who can fuse speed with discernment, automation with accountability.

 

The real AI risk ahead is how to avoid sounding like everyone else. As AI tools level the playing field in terms of being able to crank out messaging and content, those who over-delegate to machines will dilute their identity, lose sight of their customers, forget their voice, and end up contributing to the white noise instead of the bottom line. The real winners will be those who are the first to figure out what the humans ought to be doing and what the machines ought not to be doing.

 

So the question becomes less about adoption and more about orchestration. Where is human curiosity, critical thinking and judgment missing from your current process? Where are decisions being automated that deserve more human oversight? What would it cost your brand if tone drifted off-course, unnoticed? Who inside your organization has both the authority and courage to say, “This doesn’t sound like us”?

 

The companies that can answer those questions without hesitation will not just adapt to AI—they’ll humanize it. They’ll create distinctly emotive and irresistible marketing that tells stories created by humans that resonate with humans. In doing so, they’ll remind the rest of the market that progress isn’t about how fast we move, but about how wisely we decide where to go next.

 

In short, the next task ahead is to accelerate our AI learning curve. No matter how many of us would love to pretend we’re experts, we’re still just at the beginning of this wild ride.