Why Intelligence Alone Doesn’t Drive Outcomes

Artificial intelligence isn’t struggling to deliver value in marketing. It’s being blocked from doing so. That is the uncomfortable truth behind one of the biggest paradoxes in enterprise technology right now, and the numbers, for the first time, make it impossible to ignore.

McKinsey’s State of AI in 2025 survey, conducted across 1,993 respondents in 105 countries, found that 88% of organizations now use AI in at least one business function, up from 78% the year before.[1] Yet when the same survey asked about enterprise-wide financial impact, the results collapsed: only 39% reported any measurable effect on EBIT, and among those, most attributed less than 5% of total profit to AI. The organizations genuinely winning, those generating more than 5% EBIT impact and reporting significant value, numbered just 109 out of 1,993 respondents. That is approximately 6%.[1]

This is not an adoption problem. It is an execution failure.

• 88%of organizations use AI in at least one business functionMcKinsey State of AI, 2025
• 6%qualify as “AI high performers” with 5%+ EBIT impactMcKinsey State of AI, 2025
• 32%of marketing organizations report full AI integrationFeedough / MarTech, 2025

The Illusion of AI Adoption

On paper, AI adoption looks impressive. [2] According to Social Media Examiner’s 2025 AI Marketing Industry Report, which analysed responses from over 730 marketing professionals, 60% of marketers now use AI tools daily, up from 37% in 2024. A further 84% report increasing their usage over the past year. The narrative suggests a rapid transformation is underway.

But operational reality tells a different story. Only 32% of marketing organisations report full AI solution implementation, according to industry data compiled by MarTech and Feedough.[3] The average enterprise now manages more than 130 applications with overlapping functionality, while marketing technology utilisation actually plummeted from 58% in 2020 to just 33% in 2023, the lowest figure on record, according to analysis from Averi AI.[4]

“Most organizations are still trapped in the pilot loop, with scattered wins, thin deployment of agents, and only a small 6% high-performer group reporting more than 5% of EBIT attributable to AI.”— McKinsey & Company, The State of AI in 2025

What companies have built is not an AI-powered operation. It is a layer of intelligence sitting just outside the systems that actually make decisions, useful enough to justify a budget line, too disconnected to change outcomes.

Where the System Breaks

The core issue is structural. For decades, marketing technology stacks have been built on deterministic systems, CRMs, ERPs, and analytics platforms designed to answer one question with certainty: what is true? These systems rely on fixed rules, predefined logic, and strict compliance frameworks. AI operates differently. It is probabilistic. It interprets context, predicts outcomes, and suggests what should happen next.

IBT SG

When a probabilistic system tries to plug into a deterministic one, the result is not synergy, it is rejection. AI outputs do not fit neatly into rigid workflows. They either violate rules, trigger compliance concerns, or fail to meet the strict thresholds required for automated execution. McKinsey’s research reinforces this directly: the organisations that have overcome it are not those with better models, but those that fundamentally redesigned their workflows. High performers were found to be nearly three times more likely than peers to have reworked processes when deploying AI, 55% doing so, versus roughly 20% of other firms.[1]

From Intelligence to Inaction

This is where most AI strategies collapse. Organisations have invested heavily in generating insights, predictions, and recommendations. But they have not built the infrastructure to act on them.

CoSchedule’s 2025 State of Marketing AI Report, based on a survey of 1,005 marketing professionals conducted in December 2024, found that while 41.65% of marketers report most or all of their tools now carry AI features, and nearly 60% plan to increase AI spending in 2025, only 42.2% have actually integrated generative AI into active campaigns.[5] The investment is real; the integration is not.

The result is decision paralysis. Teams receive smarter insights but remain bound by legacy processes. Campaigns still require manual approvals. Personalisation remains limited by predefined segments. Real-time decisioning, the headline promise of enterprise AI, never fully materialises. Companies are not lacking intelligence. They are lacking coordinated action.

• 60%of marketers now use AI tools daily, up from 37% in 2024Social Media Examiner, 2025
• 42.2%have actually integrated generative AI into active campaignsCoSchedule, 2025
• 55%of AI high performers fundamentally redesigned workflowsMcKinsey State of AI, 2025

The Output Problem No One Talks About

Even where AI is used successfully, a different problem is emerging. The outputs are becoming indistinguishable from one another, and the research documenting this is now peer-reviewed.

AI tools are widely used but rarely connected to real decision systems
IBT SG

A July 2025 paper published on SSRN, Generative AI and Content Homogenization: The Case of Digital Marketing by Liu, Wang and Yang, studied the restaurant industry, a sector where 70% of establishments are independently owned. It found that content generated by popular large language models appears more similar to each other compared to human-created content, a phenomenon the authors term “content homogenization,” which they warn may dampen consumer engagement and dilute brand uniqueness over time.[6] A separate 2025 study in the journal Sociological Methods & Research found that LLM-generated responses are measurably more homogeneous than human-written equivalents, and that people are beginning to adopt LLM-like linguistic patterns even in their own written communication.[7]

“Generative AI is trained on the average of the internet, which means its default output sounds like everyone else’s. For brands that have spent years building a distinct voice, tone and perspective, that’s a real problem, especially at scale.”— Contentstack, Brand Voice in the AI Era, 2025

Brands are producing more content than ever, yet sounding increasingly alike. Messaging becomes polished but generic, consistent but forgettable. In trying to scale communication, companies risk diluting the very identity that differentiated them. This is not a content problem. It is, again, a systems problem, one where speed is prioritised over strategy, and volume substituted for voice.

The Real Shift: From Tools to Orchestration

Fixing this does not require better models. It requires better architecture. McKinsey’s findings are explicit on this point: the gap between AI high performers and everyone else is not technological, it is organisational. High performers are 3.6 times more likely than other firms to say they intend to use AI for transformative, enterprise-level change, rather than incremental efficiency gains.[1] They commit budget to match: more than one-third of high performers allocate over 20% of their digital budget to AI, making them five times more likely than peers to make a serious bet on the technology.[1]

The practical implication is clear. Instead of asking AI to replace existing systems, organisations need to design environments where it can interact with them, within clear guardrails, aligned with business rules, compliance requirements, and real-time context. That shift changes the role of AI from a tool that produces suggestions to a layer that enables action.

What Comes Next

The gap between AI capability and business impact is no longer a technical limitation. It is an organisational one. Companies have access to powerful models. They have data. They have clearly identified use cases. What they do not have, in most cases, is a system that allows all of these to work together in a controlled, scalable, and governed way.

Until that changes, most AI investments will continue to underdeliver. McKinsey’s own summary is blunt: “meaningful enterprise-wide bottom-line impact from the use of AI continues to be rare.”[1] The real competitive advantage in the next phase will not come from who has AI. It will come from who can actually use it to make decisions, and execute on them. Because in the end, intelligence alone does not drive outcomes. Action does.

Sources & References

1. McKinsey & Company. The State of AI in 2025: Agents, Innovation, and Transformation. McKinsey QuantumBlack, November 2025. Survey of 1,993 respondents across 105 countries.
2. Social Media Examiner. 2025 AI Marketing Industry Report. Analysis of 730+ marketing professionals, 2025.
3. Feedough / MarTech. AI Marketing Statistics 2025. feedough.com, 2025.
4. Averi AI. The State of AI in Marketing 2025 & Beyond: 7 Trends. averi.ai, December 2025. Citing Gartner martech utilisation survey data.
5. CoSchedule. State of AI in Marketing Report 2025. Survey of 1,005 marketing professionals, December 2024 – January 2025.
6. Liu, Chaoran; Wang, Tong; Yang, S. Alex. Generative AI and Content Homogenization: The Case of Digital Marketing. SSRN Working Paper, July 26, 2025. DOI: 10.2139/ssrn.5367123.
7. Zhang, Simone; Xu, Janet; Alvero, AJ. Generative AI Meets Open-Ended Survey Responses. Sociological Methods & Research, May 2025. DOI: 10.1177/00491241251327130.