The State of GenAI Adoption 2025

The advertising industry is no stranger to transformation. It tends to come in waves. Over the last roughly 30 years, we’ve witnessed three major disruptions to our ecosystem:

•     The internet, which made the world’s knowledge accessible with a keyboard and a mouse;
•     The smartphone, which put that knowledge in our pockets; and
•     Social media, which profoundly changed how people consume media, access news, and connect with each other.

Each of these waves had a dramatic and lasting impact on the nature of our business, driving massive shifts in consumer behavior, media spend, the types of services agencies provide to clients and much more. As a result, today’s advertising business is almost unrecognizable compared to 40 years ago.

Now, with AI and all its variations, we face not just a single wave of disruption, but a series of tidal waves so massive and so fast that it can be difficult to catch your breath. Navigating this change requires smart use of technology, business strategy, and change management. It also requires reliable data to evaluate and assess where competitors are in their adoption journeys, how the landscape is shifting, and what barriers remain – all of which will help inform an agile, strategic adoption approach that leads to differentiation and growth. 

We recently released our 2nd edition of our State of GenAI Adoption study, conducted in partnership with Forrester. Members can access the full report here but below are highlights and key insights.

Adoption is growing, but frequency remains nascent

We saw significant year over year growth in overall adoption, with now 75% of respondent agencies using GenAI vs. 61% in 2024. Many of those that were dabbling or exploring last year have become active users as the promise and potential benefits of GenAI become more clear. The potential benefits cited as important include drivers of business growth both for the agency and clients – aspects such as improved employee productivity (89%), improved impact of campaigns (61%) or enhanced personalization (67%). There was also a marked increase in those that see potential efficiencies and cost savings, particularly from a production standpoint – 69% of respondents highlighted this in 2025 vs. only 13% in 2024.

And yet, the vast majority of respondents (nearly 80%) report that they use GenAI on LESS THAN HALF of their work. Why? Shadow AI usage (employees using unsanctioned AI tools without disclosure)  is likely a contributor here, but just as important – many barriers to adoption remain. They can be summarized in three broad categories:

1.     There is a gap between now and next

Less than a third of respondents believe that GenAI has already impacted their business across key dimensions. But, asked to predict the ways that GenAI will impact the agency business in the next two years, respondents foresee much more significant changes. Three quarters see changes to the amount and type of content we produce for clients – as platform explosion continues and fragmentation persists, more content of different types, shapes and sizes is necessary to feed the beast; and AI-powered personalization promises to finally fulfill the 1:1 marketing vision, but requires more volume, and more agile content. 

Perhaps most importantly, nearly 70% believe that changes are coming to the agency business model and the composition of their organizations. There is clearly a recognition of the NEED to change, but what exactly that change looks like is nebulous at this stage. 

Paralyzation caused by this ambiguity risks agencies falling behind. Those that fail to proactively experiment with AI-enhanced models may face downward pricing pressure, eroding margins,  increasing competitive disadvantage and a shrinking addressable market as client expectations evolve.

2.     Fear, Uncertainty and Doubt Cloud the Path to Full Adoption

AI raises many profound and existential questions, while the surrounding legal and regulatory swirl creates significant gray area. We asked survey respondents about challenges to adoption across their agencies, and the multiple-choice question had dozens of options to select from. Even the least frequently selected options were still flagged by a quarter or more of our respondents, suggesting that AI barriers are many – and they are significant. Legal concerns remain the #1 barrier, cited by 77% of our respondents – the exact same number as 2024. Data privacy, security, accuracy, reliability and bias are close behind.

Employee readiness, lack of technical skills, and lack of understanding of GenAI tools were also popular responses – suggesting a relatively urgent need for employee upskilling. The “lack of understanding of GenAI tools” option saw a significant jump vs. 2024, more than doubling to 42%. This is not surprising in an ecosystem of more than 80,000 tools, with new startups launching seemingly every day while the entrenched players roll out new features quickly as well. Quality of AI output also remains a concern, despite breakneck advances in GenAI capabilities. 

3.     Show Me the Money

Roughly a third of our respondents reported struggles with GenAI strategy and budget, while around a quarter also raised concerns over the ROI of the technology. When asked about the anticipated impact of GenAI on company headcount and personnel budget, very few predicted  growth and the vast majority forecasted a less than 10% reduction in both. Contrast this with the nearly 70% of respondents who see efficiencies as a key potential benefit of GenAI, and the business model challenge agencies face becomes clear. Maintaining the same headcount and personnel budget in the face of AI efficiencies requires either productivity-driven revenue growth or a business model not based on hours and bodies. And again, roughly 70% are predicting changes to the business model and composition of the organization over the next two years.

To make this challenge even more daunting, when it comes to the remuneration model for GenAI technologies, fully three quarters of our respondents are treating it as a cost of business. This is not sustainable, nor is it a recipe for revenue growth. Without adjusting pricing, agencies risk devaluing their services and limiting their ability to invest further in AI innovation or attract top talent. Our respondent base seems to recognize this, with many predicting a shift to pass through or cost-plus remuneration models becoming more common by 2027. Still, less than 15% predict that a “line of business” model will be implemented within the next two years. The survey defined line of business as follows: “Generative AI capabilities are considered a line of business inside the agency and sold to client separate from traditional services.” The 4As recognizes that this is an ambitious objective, but believes this model deserves more attention and may well be a key to avoiding pricing model traps of the past. Multiple paths exist here, but one suggestion would be productizing GenAI offerings (e.g. – AI production, synthetic audiences for campaign insights, AI-powered pre-testing, generative engine optimization, etc.) and building those solutions as a combined line of business tailored to meet individual client needs.

Key implications for agencies:

•       Adoption is up overall, but frequency continues to lag due to barriers around legal, data security, employee readiness and business model concerns. These challenges should not prevent adoption, though – your competitors are rapidly adopting and GenAI usage is becoming table stakes. The only way around these barriers is to experiment responsibly and follow best practices outlined in industry resources like the 4As GenAI Blueprint.
•       Employee training is an urgent need, and is best supported by a multi-pronged approach that includes: 
  • formal training coursework (like those offered by the 4As, including our Generative AI Certification Program)
  • peer-to-peer learning programs
  • regular sharing of use case discovery, best practices, evaluations of new tools, etc.
•       The business model that got us here will not get us there. The efficiency gains unlocked by GenAI will be undeniable, and when AI is saving significant amounts of time … time is probably not the unit you want to build your pricing model around. We recommend exploration of alternative compensation models to ensure a future-proof business model. Our paper “Decoding Compensation Models & Implementing the Right Model,” written in partnership with the ANA, provides a thorough analysis of options and related implications, helping you move beyond time-based pricing to models that capture value from IP, outcomes, or creative advantage. 

Ultimately, when it comes to AI, the tech is the easy part. Leadership and change management are the hard bits. Navigating these tidal waves of AI requires strong leadership and a focused approach to change management to foster a culture of experimentation, manage risk, and redefine the agency’s value proposition for the AI era.

For a deeper look into why the business model is a key challenge for GenAI adoption, read our related article AI is a Growth Engine, Not a Cost Cutter.