The content supply chain’s AI awakening

Key takeaways

Despite initial optimism, the integration of gen AI into content supply chains (CSCs) has been slower than anticipated. Only 50% of organizations had achieved their planned AI adoption by the end of 2024, falling short of expectations.

Yet, enthusiasm is higher than ever. Almost two-thirds (64%) of organizations are more enthusiastic about gen AI’s impact on CSCs than they were a year ago. 84% still say gen AI can successfully scale the creation and delivery of personalized experiences. 

Roadblocks are common and can cause adoption to stagnate. Cost, organizational change, risk mitigation, and lack of trust in AI-generated content can stall adoption. Overcoming these challenges is necessary for organizations to realize the full benefits of an AI-powered CSC. 

Going “all in” reaps the highest rewards. When organizations take a holistic approach by investing in strategic planning, robust financial and human resources, proactive change management, among other strategies, they report—when compared to their peers—a 22% higher return on their CSC investment and a 30% higher ROI on their gen AI integration.

Introduction 

Last year, we published The revolutionary content supply chain: How generative AI supercharges creativity and productivity. The future was all rosy and all gen AI. Finally, organizations had the capability to deliver on the promise of content personalization at scale and frictionless automation, with data insights at their fingertips to direct channel strategies and investments. 

This year’s research revealed a reality check and a paradox: while the surge of enthusiasm around gen AI and content supply chains (CSCs) has dramatically increased—and real progress has been made—the speed of operationalization has fallen short of organizations’ optimistic expectations.
 

A content supply chain brings together people, processes, and technology to effectively plan, create, produce, launch, measure, and manage content.

Case in point: last year, 74% of executives reported their organizations were either already using or planning to use gen AI by the end of 2024 for numerous CSC use cases, from content creation to automated workflows. Yet today, only 50% have achieved this milestone.

Initial expectations for such rapid adoption were not surprising, given gen AI’s early hype cycle when it exploded into the public domain just over two years ago. But the figure below illustrates examples in which reality has fallen short of projections: some of the most popular CSC/gen AI use cases have taken longer to implement than expected. 
 

For some CSC use cases, gen AI adoption is trailing expectations

Q. What types of research and analysis tasks are you already using gen AI for? If you’re not using gen AI in these areas today, when do you plan to do so? Questions asked about customer insights, creative ideation/conception, media mix/budget planning, and channel optimization.

Why?

Call it the school of hard knocks: experimentation and early pilots uncovered complexities that organizations need to address before they realize the full benefits of an efficient CSC. With incessant pressure to gain competitive advantage and satisfy consumer demands for personalized experiences, companies are scrambling over the learning curve to master these challenges.

To better understand where organizations are struggling most and how they are recalibrating their trajectory forward, the IBM Institute for Business Value (IBM IBV) partnered with Adobe and Amazon Web Services (AWS) to do a deeper dive into the status of organizations’ CSC journeys. With Oxford Economics, we surveyed 1,100 executives from major enterprises with an average of $22 billion in annual revenues. Our sample includes organizations from 12 industries across 15 countries to arrive at a comprehensive look at how large organizations are currently using gen AI to support their content-related activities. 

We learned there is no single, prevailing CSC challenge, but rather several important barriers—some of which were previously underestimated—that are slowing implementation and adoption of gen AI-fueled CSCs: costs, organizational impact, risk mitigation, and a lack of trust in AI-generated content.
 

But good news: Organizations are still realizing a number of benefits, including reduced production times, enhanced content creativity, and increased content flexibility while mitigating the very concerns that are slowing implementation for many (see figure). 

Large percentages of leading organizations report their gen AI investments are positively impacting their CSCs 

The motivation to advance is clear. Even while wrestling with challenges, almost two-thirds (64%) of organizations are more enthusiastic about gen AI’s impact on CSCs than they were a year ago. Indeed, 84% still say gen AI can successfully scale the creation and delivery of personalized experiences. The will is there—what organizations need is a path forward. 
 

Part one

Navigating the challenges: Advancing gen AI in content supply chains

Demand for content is accelerating and the percentage of revenue devoted to its creation and delivery is increasing. In the last year alone, surveyed executives estimate having spent, on average, $167.7 million on their content activities. By 2026, they expect that percentage to jump 9.7% to $184 million. 

It will not be possible to meet this growing demand for content without CSCs that are driven by gen AI capabilities, which is why these solutions have energized content teams. Organizations have begun integrating AI into their CSCs, but their progress has been 32% slower than anticipated across 30 distinct CSC use cases.

When we asked organizations what impedes their progress, they cited a number of challenges (see the figure below). Of the various obstacles faced, these were the most prevalent:

1. Cost. Based on our 2023 CSC survey, nearly two-thirds of organizations felt the budgets needed to cover CSC expenditures were adequate. A year ago, “difficulty obtaining required funding and budget approvals” lagged well behind other CSC barriers that organizations were facing. Fast-forward 12 months and cost surfaces as the top obstacle most executives (61%) report as hampering the integration of gen AI into their CSCs. To some extent, this is a consequence of organizations jumping feetfirst into gen AI experimentation with easily accessible public platforms. But content creators quickly realized that these models required refinement and the addition of proprietary data to meet their needs. 

Early pilots also revealed data and infrastructure vulnerabilities, the possible need for hardware and software upgrades, additional licensing, and the hiring and upskilling of new talent—all of which add costs. In 2023, nearly half of businesses conducted initial experiments in isolated pockets of the business with few guardrails. Without orchestrated, strategic directives, many early pilots missed the mark, making it difficult to build a concrete business case for budget-conscious organizations. 

“Some of our early experiments made us think very carefully about what’s the right amount of data, with the right quality, at the right time, with the right auditability,” observes James O’Keefe, Marketing Technology Director at UK-based Legal & General. He adds that being selective about data can minimize cost. “It’s about trying to find the right use cases that will drive value and attract budget.”
 

2. Organizational change. Adapting to new technologies and processes can spark an evolution—or revolution—in workforce skills, job roles, and company culture. More than half of executives (57%) say organizational change is a critical challenge impacting the improvement of their CSCs, and they claim it is the most difficult barrier to overcome. 
 

As Chris Muscutt, Director of Content Supply Chain/Marketing Technology at Philip Morris International (PMI) noted, “People will go off and do gen AI with or without you. The trick is, how do I align those projects that went ahead without me, so they still drive an aligned agenda?”

An end-to-end CSC can touch multiple disciplines, departments, teams, and locations. Data and content assets may reside in a variety of formats and on platforms that aren’t integrated. For an enterprise CSC to be effective, organizations need to break down data silos and align not only technologies and workflows, but also priorities and strategic agendas, requirements, budgets, skillsets, timelines, metrics, and expectations. But resistance to change, fear of job displacement, and lack of a common vision and objectives can derail even the best-laid plans. 

However, challenges can also present opportunities, notes O’Keefe of Legal & General: “Our CSC helped us see how—as an organization—we can collaborate across different teams and departments. We combined some gen AI/CSC experiments because it was inefficient for multiple teams to be exploring similar things. So, we gathered a ‘coalition of the willing’ into a steering group and working groups. We went first and broke the ice so others could fast follow.”
 

3. Risk mitigation. When considering the impact of integrating gen AI into their CSCs, 57% of executives expressed concerns about insufficient risk controls. Organizations must navigate potential pitfalls associated with AI, such as data privacy breaches, algorithmic bias, and the ethical implications of AI-generated content. Because risk management is a key component to addressing any challenge, we take a closer look at this in the next section.
 

4. Lack of trust in AI-generated content. In addition to concerns about cost and risk, 56% of executives also fundamentally worry that AI-generated content might backfire on them. Given the nuanced nature of human language and the potential for AI to misinterpret or misrepresent information, the accuracy and reliability of AI-generated content—especially without appropriate governance—can present significant challenges. 
 

The most concerning risks of incorporating gen AI into CSCs, by percentage of organizations citing the factor.

Optimizing ROI through a holistic, gen AI-focused approach to strategic CSC investments  

As the initial AI fervor inevitably gives way to getting-it-done realism, top companies are demonstrating that strategically investing in CSC workflows, leveraging gen AI assistants, integrating robust governance practices, among other strategies, can yield substantial returns. We’ve identified these top companies as those who are further along with their gen AI/CSC integration and seeing the highest positive business impact. 

Instead of pursuing one-off, isolated initiatives, advanced enterprises allocate more resources toward streamlining processes, automating repetitive tasks, and enhancing interdepartmental collaboration, which enables them to reap the benefits of a more agile, responsive CSC. And, as our research has made clear, top companies are keenly motivated by business results. 51% are investing in CSCs with the goal of increasing revenue, compared to just 35% of their peers.
 

Leading organizations not only navigate the challenges of AI integration more effectively but are also achieving impressive results, with an overall ROI on their CSCs that is 22% higher than their peers and, more specifically, a 30% higher ROI for their gen AI initiatives.

However, merely investing in advanced technology is not enough—leading organizations engage in proactive practices (see the figure below). Organizations taking a comprehensive approach also understand the importance of integrating strong governance practices. They implement robust frameworks to help ensure AI's ethical use, manage risks effectively, and build trust in AI-generated content. This includes establishing clear guidelines for AI applications, conducting regular audits, and maintaining transparency in AI operations. Such practices not only mitigate potential problems but also enhance stakeholder confidence in the organization's AI capabilities.
 

Leading organizations are more likely than their peers to adopt these practices

 
Perspective: 

Part two.

Managing risk: Unleashing creativity with gen AI 

Vigorous risk management is essential for realizing the full creative potential of gen AI. Companies taking a holistic approach to gen AI have mastered this art, effectively managing risks in a number of areas. And by doing so, they are able to embrace gen AI deeply and elevate creativity within their organizations. 

Data provenance, the digital trail that records the origin of data, is a critical risk area for gen AI. Leading organizations report extensive use of this approach—67% more than other enterprises. By helping ensure data accuracy, integrity, and traceability, they give their AI systems street credibility. Vigorous data management empowers confidence in AI to generate reliable, valid outputs that provide a launchpad for creative exploration and innovation.

To that end, O’Keefe of Legal & General notes, “We had some upstream data sources that were still in silos, so we had to focus our early efforts to get to a single, aggregated customer view. We needed to have more people using a consistent standard and a consistent DAM [digital asset management system]. In effect, we had to rethink the use of the DAM. The notion of DAM being a strategic data asset is important.”

Trust in AI-generated content is an area that can significantly impact trust in an organization’s brand. In fact, when compared to enterprises overall, advanced organizations report a significant improvement in this area—46% more than their peers. How? They implement transparent processes and validate AI-generated content against formal standards. As a result, they report extensive use of ongoing model training and refinement 48% more than their peers. This high level of trust empowers teams to embrace AI-generated content, fostering collaboration between humans and AI. 

Additionally, scrupulous oversight of gen AI outputs helps ensure greater accuracy for leading organizations, as well as adherence to brand standards, with 18% more of them taking a comprehensive approach to gen AI by exercising this level of rigorous oversight. They use AI as a tool to augment human creativity rather than replace it. By assigning laborious, repetitive tasks to AI, these organizations free human creators to focus on the more strategic aspects of their work. And because AI outputs are programmed to be brand-aligned, content creators can trust AI to jump-start ideation, accelerating the creative process. 

O’Keefe of Legal & General notes the importance of quality control and compliance to build trust in gen AI outputs: “We constantly check the validity and quality of the gen AI models we deploy. We started by running low-risk opportunities that could be used in a fairly controlled manner. Now we run phased deployments for a particular audience that’s exposed to content over a period of time, and we monitor the outcome. This affords us higher levels of quality control and helps build confidence that we’re achieving the requisite compliance standards.”

The result of these effective risk management strategies? With data management and trust verification under control, human creators are free to innovate boldly and take calculated risks (see the figure below). The combination of advanced AI capabilities and human ingenuity creates an incubator for ideas and solutions.
 

How advanced organizations use risk management to foster innovation

Organizations across the board say gen AI has boosted their creativity. The majority of executives report that their gen AI-powered content supply chain investments have had a positive impact on their creative process (see the figure below).

Betsy Rohtbart, VP of ibm.com, observes, “AI’s first ideas may be nonsensical or awesome, but either way, it stretches the boundaries of our creative process. It helps us rise above groupthink. And it’s created a comfort zone for the greater sharing of ideas, because when a concept comes from AI, not an individual, the team feels freer to judge it objectively. We’re able to get a first cut developed quickly. Then we pick up the pen to make sure human emotion is reflected.” 

And Chris Muscutt of PMI adds: “The challenge [around releasing creativity] has been around creating clear governance language of what you can and can't do with the new tools. Now that governance is in place, people feel more free, more confident to experiment.”
 

Impact of CSC investments on creativity across all organizations

Part three.

The next wave of CSCs: Small language models and agentic AI

Once a niche concept, small language models (SLMs) are now becoming a foundational component in CSC strategies. SLMs are trained on specific, targeted groups of data that generate customized output. One huge upside: data is kept inside an organization’s firewall, so sensitive data is not leaked to external SLMs. With more modest hardware and training costs, SLMs can reduce the cost of ownership compared to large language models (LLMs). They’re poised to revolutionize personalized and scalable content creation.
 

A mere 12 months ago, organizations were mostly looking to LLMs to support their CSCs. Today, 41% of organizations have already pivoted to integrate SLMs into their operations, and an impressive 87% are planning to do so by the end of 2025. This surge in adoption attests to SLMs' potential for enhancing efficiency, personalization, and scalability in content creation. In fact, 60% of organizations using SLMs today are seeing a positive impact on the operating costs of content.

As well, the fact that SLMs are developed for a particular purpose and can be more secure means they can be more trustworthy—a huge benefit given that almost six out of 10 executives say lack of trust in AI-generated content is a major hurdle. 

The trend toward smaller, more efficient models has its nuances. For example, not all SLMs are proprietary, but increasingly, organizations are indeed opting for proprietary models. This shift reflects a broader strategy, moving away from one-size-fits-all solutions toward more targeted, bespoke approaches to language modeling. 

Proprietary or not, SLMs signify a paradigm shift in the approach to content creation. Their compact size and efficiency make them ideal for edge computing and mobile applications, opening up new possibilities for personalized and contextually relevant content delivery.

Leveraging gen AI assistants, or agentic AI, can be another victory. Unlike traditional AI tools, agentic AI can create content, predict outcomes, and suggest optimal solutions, thereby augmenting human capabilities and enhancing human productivity. In other words, they are autonomous systems that can execute complex workflows (see Perspective “Agentic AI—The content creation revolutionizer”). Leading organizations have harnessed this power to accelerate content generation, reduce production times, and improve content quality.

Perspective 

Action guide

How can organizations realize the promise of an intelligent CSC without getting sidelined by challenges with data, processes, and/or adoption? The following actions, informed by our data-driven insights, provide a path that can accelerate your CSC/gen AI journey.

1. Go big, but with focus.
By definition, pilots are small in scope, which is useful for agility and experimentation. But once you’re committed to implementation, you need to capture a holistic view of CSC and gen AI requirements to identify gaps, minimize redundancies, and improve efficiencies. However, when transitioning from a pilot mentality to an implementation strategy, avoid the temptation to “boil the ocean.” 

Steps to take:

• Note interdependencies among departments, systems, and processes. 
  For those businesses that have been piloting gen AI CSC use cases concurrently across isolated silos, now is the time to audit these activities and their outcomes. Beyond those who are piloting, think broadly to uncover who else your CSC touches and how. Then identify their technical, organizational, and strategic requirements as well.
• Prioritize a select group of mission-critical gen AI use cases. 
  Armed with a comprehensive map of your organization’s CSC needs, note which early pilots resulted in quick wins and which revealed where more effort and investment are needed. Weigh the costs and benefits: which use cases will deliver the biggest returns, and which ones are foundational and need to be in place before others can be implemented? Then chart your course in phases.
• Define and measure your baseline and success metrics. 
  Let your CSC’s North Star guide you toward what needs to be measured. Different stakeholders may have specific CSC goals, key performance indicators (KPIs), and metrics that you’ll need to reconcile. To keep stakeholders engaged and supportive, report progress consistently, be transparent about the areas that need more attention, and proactively identify next steps that build on past achievements. 

“It’s important to strike a balance between the pressure to move quickly and the pressure to get it right.” 

James O’Keefe
Marketing Technology Director, UK-based Legal & General

2. Don’t shortchange change management.
No matter how smart your strategy, if people aren’t convinced it serves them in some way, their reluctance will slow down your progress. Worse, it could sink your initiatives. Yet, organizations often focus on the tactical aspects of transformation, underestimating the importance of rallying their teams. 

Steps to take: 

• Create a cross-functional AI adoption approach that aligns people from marketing, content development, IT, and operations teams. 
  Having representation from each of these areas enables a cross-pollination of perspectives to generate awareness of the concerns of a variety of stakeholders.
• Identify advocates to lead the charge. 
  Passion is infectious. When people see others eagerly adopting new technologies, tools, or processes, and the benefits are clearly demonstrated, curiosity and enthusiasm grows.
• Place people at the center of your CSC cycle of creation, delivery, and measurement. 
  Your AI proposition is ultimately about empowering people. Carefully plan for levels of engagement that are targeted to meet the explicit needs of specific CSC hero-users. Involve your hero-users from the beginning, in the experience design process as well as refinement and training. Enable a responsive feedback loop so that individuals can easily get the support they need, feel heard, and start the mental and emotional shift of buying-in and playing an active, essential role in your transformation’s success. 

“Create and optimize templated ways of working that streamline workflows, avoid duplication, and better enable automation.”

Chris Muscutt
Director of Content Supply Chain/Marketing Technology at Philip Morris International (PMI)

3. Free creative capacity by mitigating risk.
When you focus your initial gen AI use cases on repetitive tasks such as  content tagging, SEO optimization, and compliance checks—and you embed quality checkpoints into workflows—you create an environment where teams can take creative risks with more confidence. Combining AI’s efficiency with human creativity enables organizations to scale content production without losing emotional resonance or strategic alignment with the business. To do this effectively, be aware that automation just makes flaws happen faster unless you’ve done your due diligence to help ensure your data and processes can be trusted. 

Steps to take:

• Develop and maintain a robust enterprise taxonomy. 
  The devil’s in the details, and tags and metadata are among content providers’ biggest headaches, made especially troublesome when teams are not aligned with common standards. Gen AI can be a huge time-saver for these users, but only if there is an underlying taxonomy to depend on. The good news: organizations can consider this a long-overdue housecleaning that gen AI can, in fact, help facilitate. A well-organized data taxonomy means content creators can use AI to generate outputs that are far less likely to present misinformation or insufficient results.
• Foster collaboration between humans and AI. 
  For people to view AI as an enabler of human creativity rather than a replacement, nothing will be more convincing than hands-on experience. Providing easy access in a safe environment to learn is a necessary first step. Real progress can be made quickly by crowdsourcing insights from users who are encouraged to apply gen AI to their processes and share the leading practices that emerge.
• Encourage (strongly) the CMO and CIO to become best friends. 
  While a CMO may be responsible for the vision for their CSC, it’s the CIO who enables the platform and technology to achieve it. Mitigating risks associated with gen AI—whether security breaches or content hallucinations—requires a strong, strategic alliance between marketing and IT. This partnership, when chartered with common goals, can seek innovations that serve both marketing’s and IT’s agendas while delivering meaningful business results.

“I’ve been able to positively redeploy and widen scope for people using AI making us a higher-scale and more efficient organization. If AI can help people be more productive, that’s great. I want my teams to be ambitious about using AI. Not reckless.”

Betsy Rohtbart
VP, ibm.com