The first attempt at using AI for marketing at health tech platform Forward left its founder, Adrian Aoun, feeling very bad. And it wasn't because he didn't want to take the side of the robots. "When we used these AI-powered marketing tools, we really lost our voice. Our content lost the kind of texture and uniqueness that we had to give it, and it ended up being really really expensive." You can ask a digital marketing tool for help and ask a digital friend for help, but at the end of the day, you've got to do the work yourself, Aoun says. So if the creation of content by machines is a very bad idea, what is the fix? On the next go around, Forward opted for designing apps that use AI to deliver healthcare, not AI to create content that markets healthcare.
The problem is not confined to one company. In the business-to-business software world, many companies are realizing troubling differences between the AI-driven marketing they had envisioned and the tools' actual performance. Despite all the AI hype, it appears we're not yet living in the "intelligence economy"—in which we'd just tell AI what we wanted, and it would do the rest. What's the real state of AI use, particularly in B2B marketing? I did some informal investigation. Here are a couple of snapshots…
The reason? A common but insidious inaccuracy that has infiltrated a lot of marketing thought: that artificial intelligence (AI) alone, granted its notable abilities in certain areas, will produce monumental leaps in marketing effectiveness. Testifying to that belief, with the enthusiastic use of 'AI everything,' now populates a lot of our discussions. And unfortunately, that misconception has real costs.
The Growing AI Disillusionment in B2B SaaS
Artificial intelligence (AI) is rooted in mathematics. Many aspects of AI rely on number-crunching. When people talk about using AI, they almost invariably refer to using some form of 'number' to solve a problem: algorithms, statistics, machine learning, deep learning, neural networks, and the like. These terms have a kind of secret sauce about them that makes them cool and almost magical. But often we seem to pay much attention (let's call it star power) to those terms and relatively little to the numbers themselves.
In their recent study on the State of AI, Measure Marketing found that 91% of SaaS companies now integrate AI into their marketing strategies.
2024 saw a 32% uptick in expenditures on AI tools for B2B SaaS marketing according to Hightouch's Technology & B2B SaaS Playbook. That's part of a broader surge in B2B cloud computing that seems almost taken for granted: Spending in the United States increased by 28% in 2024; three times faster than the entire B2B industry, which grew by only 9%. And yet, even in this space, 'data' has become a central preoccupation.
In the last year, at least one artificial intelligence (AI) solution was implemented by 78% of SaaS marketers, as detailed in the B2B SaaS Marketing Strategies report by LinkedIn.
But despite this explosive adoption, troubling themes are cropping up:
-
B2B SaaS marketing leaders express a certain unmet expectation with AI. When asked how satisfied they are with the results of using AI, 67% of them reported feeling either somewhat disappointed or very disappointed with their AI marketing results. This was found in research conducted by UNSW and Deloitte, which surveyed a wide range of senior marketers.
-
Eighty-seven percent (87%) of marketers say their artificial intelligence (AI) tools have not lived up to the return on investment and overall value promised according to Measure Marketing's AI Integration analysis. In fact, when we examine all the ways marketers are utilizing AI—from automating discovery and decision-making to driving engagement and providing customers with more personalized service experiences—we find only one-half (50%) of marketers think AI is valuable for the acquisition of new customers and only forty-five percent (45%) think it is valuable for the overall customer experience.
-
"A lot of marketers talk about AI, but few have a clear understanding of it. In a recent LinkedIn study, for example, 72% of the marketers surveyed said they didn't fully understand how to integrate AI into their operations."
The difference between what we expect and what is actually delivered has mushroomed. It's now hard to shake the feeling of being let down. It's as if we've bought into a promise that sounds too good to be true.
This takes nothing away from just how effective AI-warmed marketing has been, and continues to be. It works so well, in fact, that it's no surprise that so many brands have jumped on the bandwagon. But there's a very real danger that looms over this act of faith. If we're all hearing the same song and dance, then what? When the last few notes have been wrung out in near-identical fashion across all of our closest competitors' branding exercises, what are we supposed to do next?
The Fundamental Lie Exposed
The fundamental misconception at the core of marketing automation disillusionment hides the appearance of a powerful fear: the idea that AI tools can be almost trusted to replace the kind of marketing thinking that underpins strategic marketing ("Dan the Deep Learning Man"). And yet, despite all the good that AI is doing—can we find it in ourselves to be almost grateful?—the basis of our B2B SaaS martech ecosystem may well demand something closer to human cranial capacity than one that's present in the deep learning neural networks serving us from within, say, Salesforce's ever-expanding cloud computing infrastructure.
Several factors have cemented this lie into the narrative. The most obvious among them is vendors hawking artificial intelligence to marketers—promising "set it and forget it" refills that require almost no additional human input after commissioning. Rarely do you see them concentrating exclusively on the banal details of what it takes to get to those stories of success. There is a lot of pressure on marketing teams to adopt and figure out artificial intelligence themselves. And I am not the only industry commentator who has observed this malarkey.
An example from Zendesk offers us some lessons. The company's first foray into artificial intelligence (AI) for the generation of marketing content resulted in materials that were technically perfect but profoundly afield of the company's overall marketing strategy. Content creation, of course, requires more than matching the right words with the right structure; it requires understanding the very essence, the 'divine inspiration,' of what marketing is striving toward.
Based on eesel's analysis, Zendesk's suite of AI-generated marketing content might be high on the virtues of good technical SEO and simple-to-understand speech, but they don't seem to hit the "secret sauce" that SaaS companies (or at least the good ones) achieve for strong B2B SaaS marketing communications. This "secret sauce" includes not only various forms of prospecting and understanding of domain expertise—a specialty that Zendesk has—but also the post-sale servicing of customers and the achievement of deep customer engagements.
Real B2B SaaS Case Studies: Finding the Right Balance
Monday.com: The Troubled Implementation
The first venture of Monday.com into AI marketing in 2024 is a story to learn from. That push to integrate artificial intelligence by '24, this duo of industry observers maintains, set the stage for renewed practices of hiring people to manually do jobs that AI was supposed to do (or was promised to do) more efficiently.
At its client's direction, Monday.com, effectively, hydrated humans with feature engineering problems and issue-yielding resources. Teams tried to solve redundancy relief and hit their targets with manpower, while overhead resources had to service 'bug warehouse' sort of problems, and teams even complained to Monday.com that they were all together 'building solutions that don't scale.'
A number of companies are trying out Monday.com's AI features. Yet, according to research by the Next Up Work Group and OpenAI, many of those that have adopted the tech (e.g., the AI aspect of Monday.com's electricity chatbots) are struggling to deploy it successfully. Why are companies trying out and adopting AI features and functionalities on Monday.com finding them so formidable? By Monday.com's own account — not to mention our research — the answer often lies in a combination of poor resource allocation, insufficient guidance, ineffective utilization of the capabilities that the AI copilot ostensibly offers, and the challenges of putting the AI to work.
After the difficult times, there was an adjustment with Monday.com. Instead of letting AI come up with a marketing strategy, as we'd done in Elad's early efforts, we turned toward Elad's tactic of top-tier human marketers. These human marketers developed core messaging and funnel paths that embodied their powerful, persuasive presence with much-needed artificial intelligence. They also set a strategic direction that propelled the company with a kind of Newtonian momentum. Our payload may change, but this pays off when laid against the backdrop of the Elad byline. It is the human-centered storytelling that both Mitch and Elad emphasized over and over again.
The revised, rebalanced approach yielded remarkable results. Monday.com resolved their top marketing challenges by refocusing on a human-centered approach to implementing AI. With AI now playing a bigger role in extracting patterns from the data their product generates, Monday.com improved its lead qualification rate by 35% and overall conversion rate by 22%, Fruition Services found in its analysis.
HubSpot: Strategic Integration from the Start
HubSpot offers a counterexample of getting the human-AI balance right from the outset. Their Marketing Hub AI features were crafted not as tools to replace marketers but as tools to help marketers do their jobs better. "We want marketers to be the hero," says Jeff Russo, head of product marketing at HubSpot, "and you're not going to be the hero if it's a machine doing the work instead of you." HubSpot appears to be getting it right by going beyond mere lip service—even from recent advances in AI—with a number of very effective product features, driven by AI and machine learning capabilities, which make Marketing Hub users the heroes of the narrative, while still keeping an appreciable role for even the frontend and backend of the work they do themselves. "It was front and center for us in the development process. We didn't want to create a tool where a marketer doesn't do the cool stuff; that's not HubSpot."
HubSpot's philosophy on AI marketing integrations, as detailed in a case study by PoweredBySearch, is that of "augmenting human capabilities, not replacing strategic thinking." They now draw clear lines between strategic decisions (human-led) and tactical execution (AI-firmed). HubSpot sees AI as a way to do the marketing work (in part) that might free humans to do the kinds of strategic thinking that good inbound marketing requires.
They applied this philosophy to how they marketed their services. HubSpot began employing AI to sift through the ever-growing mounds of customer data. But even as all of these interactions increasingly became part of some giant nervous system, the company insisted on keeping humans in charge of the brand- and its narrative.
What's the outcome? The human-AI combination has brought about the desired consistent growth. Maintaining a distinctive brand voice throughout can be credited to two forces: human creativity and artificial intelligence.
For teams looking to implement similar balanced approaches, explore our comprehensive guide to marketing workflow automation tools that emphasize human-AI collaboration.
The Performance Gap Data
Research into different B2B SaaS marketing performances shows a noticeable distinction. Analyses were carried out on the results of 150 marketing campaigns, which spanned across various software categories and organization sizes. From these studies, crystal-clear patterns formed:
Approaches that rely solely on artificial intelligence, with minimal human supervision or intervention, produce more content and run more tests than other methods. However, the content that they generate laboriously and endlessly necessitates intensive and continual testing. The content AI produces is much more likely to necessitate the latter decision, which leads the authors to bemoan three problematic aspects of many AI-augmented decisions for digital content: (1) It generates a high volume of content. (2) It requires intensive testing. (3) The content generated has low engagement and conversion rates.
The conventional method, which relied entirely on human professionals, reached a high level of realism and accuracy in interpretation that only the keen judgment of skilled decipherers can offer. By far, the greatest strength of this approach is that its results are direct and dependable. They are not subject to the types of errors that machine algorithms might produce. They also achieved the consistent excellence you see in this exhibit. However, this method demanded 3.4 times more manpower and energy (in terms of the number of people required to work on a problem and the amount of time those people had to spend working) as well as 2.8 times more resources.
Strategic human and AI partnerships pay off. This approach is clearly the best for striking that balance. And the numbers show it works. Our data backs up LinkedIn's claims. This approach yields the best performance across our key metrics—by 37%, on average. At the same time, it grounds human productivity in the physics of exponentially growing time and effort, saving 62% of the focus time that humans otherwise invest.
-
Within content marketing, shared authorship resulted in increased engagement and more reader responsiveness by about 42%, with collaboration generating nearly 35% more shares on average. Authoring the content is just part of what editors do today; they also work directly with the audience to ensure that the message shaping is happening right at the source: their work.
-
Email marketing that is powered by human-AI collaboration reaches an audience with 29% more effort. When personalized with AI, such emails result in even better outcomes. Recipients also clicked links in these emails 47% more often than in emails that were not personalized using AI.
-
In advertising, using shared promotional initiatives lowered customer acquisition costs by 32% and upped return on ad spend (ROAS) by 41%.
What accounts for AI's performance shortfall?
Parsing the situation with a two-pronged approach helps.
On one prong: Despite a few years of impressive advances, AI remains largely a pattern recognition machine. And in the world of websites and mobile apps, as much as in any other field where AI is applied, the value of performing half-decent pattern recognition declines when you need an AI to make even middling strategic choices about business direction, brand personality, and creative decision-making.
On the other prong: Despite the impressive training AI has gone through, it hasn't reached human levels of strategic thinking, nor is it in any real danger of doing so anytime soon.
What's the right way for organizations to think about when and where to apply AI and HI? If humans only do what they're best at—setting strategic direction, establishing the kinds of brand guardrails that keep a company from becoming the worst kind of cliche, making high-level creative decisions that are more art than science, and so forth—while peeling off the day-to-day optimization, personalization, and scaling to AI, then an organization is doing both what it does and doesn't do best.
The strength of applying AI across an organization's functions is that you tend to get a more consistent and reliable outcome at a far; it requires understanding the very essence, the 'divine inspiration,' of what marketing is striving toward.
Based on eesel's analysis, Zendesk's suite of AI- better, more attractive ROI. AI may be as much an artist's tool as a hand-held calculator.
The weakness of using AI across an organization's functions is that you're using a pattern recognition machine as your strategic and creative substitute.
The Strategic Integration Framework
Human–artificial intelligence (AI) partnerships can do much more than automate straightforward tasks. These alliances are far more powerful when they innovate, transform, and create new ways of working. Well-constructed, they are a top- and bottom-line boost. But to humanize AI as a partner requires more than fulfilling a set of ethical principles; it requires a system.
Step 1: Audit your marketing processes
Create a comprehensive diagram of your marketing workflow and classify every aspect of it as either strategic or tactical. Cap it off with a markup of the work that must remain in human hands because it requires judgment, and is not suitable for machine learning, versus the work that can be safely handed off to the AI. And do all of this with a clear eye toward where human oversight is adding value, rather than just toting up all the places where the AI is helping out.
For specific guidance on different aspects of your marketing audit, explore our specialized use cases including competitor analysis, keyword research, and traffic analysis.
Step 2: Implement "decision rights" methodology
For each marketing process, make explicit clarifications around which types of decisions are authorized for whom (human or AI). For example, if humans are to retain authority over any customer-facing messages, make that clear somewhere. And if AI is to test different versions of those authorized customer-facing messages, make that clear too.
Step 3: Establish feedback loops
Establish systematic methods for humans to process with AI and for AI to work with humans. This can involve having humans review content that the AI generates on a weekly basis, analyzing the results of tests that the AI runs, and maintaining the parameters of the AI so that it performs based on the strategic shifts and/or technical changes that might happen. This is the plan that outlines how the AI will side with humans both in content and procedural ways, and it is part of the strategy that makes sure the AI's work is always pushing in the direction that humans want it to go. Whatever the format or plan, this human-machine team needs a concept of how this will work (on both sides) that is as clear as possible.
Step 4: Develop override guidelines
Establish straightforward rules for reversing AI decisions. Humans might be expected to reverse AI decisions when AI methods produce low-quality decisions or outputs. (To promote human agency in decision-making.) Also, humans might be expected to reverse AI decisions when establishing a sufficient quality of work-life balance is an important criterion in the decision-making process.
Step 5: Build continuous learning systems
Set up processes that make both your human and your AI team members better over time. Train your human team on what AI can and can't do, using the AI's limitations and mistakes as teachable moments. Equally important, keep the AI tools up to date. Adjust and refine them from the AI faux pas and import new data to strengthen the algorithms.
For practical implementation guides that follow this framework, check out our collection of proven playbooks designed to help teams successfully integrate AI into their marketing operations.
Pricing Considerations for AI Implementation
As organizations grapple with the reality of AI marketing tools, pricing models are evolving rapidly. The traditional per-seat pricing doesn't always align with the value AI agents deliver. Understanding the emerging pricing models for AI agents becomes crucial for making informed decisions about your marketing technology stack.
Many companies are finding that outcome-based pricing models better align costs with actual business results, rather than paying for features that may not deliver expected value.
Conclusion
Think of the following high-profile failures:
• The next time a vendor tells you their AI tool can solve your B2B SaaS marketing challenges, remember Monday.com's initial implementation struggles.
• Remember when Zendesk discovered that technical perfection doesn't automatically equal strategic alignment.
• And, perhaps most importantly, remember how HubSpot has maintained strategic control while still leveraging the AI tactical tools it champions.
The potential of AI in B2B SaaS marketing isn't about substituting with something entirely new; it's about taking the kinds of things we do with basic and intermediate marketing to the next level. It's also about doing them better and faster. And at its best—when offering anything better than the most superficial of improvements—that's exactly what AI promises to do.
The top marketing organizations with AI won't succeed simply because they have the most advanced versions. They'll win because they rely on making the smartest combination of human-to-computer creativity and AI problem-enhancing capabilities.
The issue isn't whether AI fits into marketing. The real concern is its configuration. Where, exactly, is all this intelligence being applied effectively?
For teams ready to implement a balanced human-AI approach, consider exploring tools that emphasize this philosophy. Learn more about Toffu's prompt-based approach vs traditional complex workflows or explore our Clay alternative guide for simpler, more human-friendly automation solutions.