The Solow Paradox gives us clues about how UK Media companies should adopt AI
Why history suggests UK media and creative agencies have time to adapt to AI — but not to waste it
This article was written by Kieron McCann Co-founder of the consultancy A few Good People
The Solow Paradox, named after Nobel Laureate Robert Solow in 1987, observed that “You can see the computer age everywhere but in the productivity statistics”. Despite massive investment in computer technology during the 1970s and 1980s, worker productivity only increased by about 1%. This disconnect between technological investment and productivity offers useful insights for media and creative agencies navigating AI adoption today.
The Evolution of the Creative Industry
The advertising industry has not been slow to adopt new technology, from computers replacing foam boards to the internet enabling digital advertising and social media creating entirely new channels. However, this technological progress coincided with a fundamental shift in business models. Through the 1960s to the early 1980s, independent agencies influenced boardroom decisions, funded by 15% media commissions. By the mid-1980s, pressure from public company acquisitions, client price demands, and diversifying promotion channels forced a shift to time-and-materials models.
This transformation moved the industry from charging for commissions to billing for production time. In the process, this gradually shifted client relationships from boardrooms to operational teams. The internet further disrupted the value chain, with Google and Meta building walled gardens that tool increasing share of advertising profits. Now AI threatens to reinvent not just channels, but the core operating model and business fundamentals of the entire industry.
Understanding the Productivity Paradox
Four decades of academic research have identified micro-level effects and macro-level effects that explain why productivity gains lag technological investment. Micro-level factors include measurement issues, in which productivity metrics fail to capture intangible benefits such as improved quality and employee satisfaction. There are also significant time lags - the 1990s productivity surge reflected 1970s-80s computer investments. Organisations needed time to adapt processes, navigate regulations, and eliminate bottlenecks in non-computerised functions. Additionally, price effects created paradoxes: Increased efficiency drove price competition, which eroded productivity gains even as product quality improved.
Macro-level factors show that the benefits of technology are unevenly distributed. Early adopters capture market share without necessarily growing the overall market, concentrating benefits among a few firms at the expense of adjacent industries. Commoditisation effects mean widespread technologies like cloud computing become utilities—everyone gains access, eliminating competitive advantage. Labour displacement occurs as innovative companies eliminate jobs, pushing workers to less productive organisations and dampening overall productivity. Geographic displacement due to globalisation has shifted production to cheaper labour markets.
The Adoption Lag is Real
Technology adoption has followed predictable patterns, regardless of the technology: internet (1990s), mobile internet (early 2000s), cloud computing (late 2000s), and social media (late 2000s). Each follows an S-curve of early hype, disillusionment, then rapid growth, followed by a plateau or decline. AI will be no different.
In the case of technology adoption in the UK, economic data have mirrored US patterns, with one critical difference: our adoption tends to be slower. Both the UK and the US saw no growth in computing productivity in the 1980s, followed by surges in the 1990s. However, by the 2000s the UK began to lag for several reasons, but one of the largest factors was regulatory constraints. In the case of e-commerce, UK planning regulations aimed at preserving high streets delayed retail e-commerce adoption compared to US companies, which rapidly built integrated online-offline distribution models.
Looking more specifically at marketing technology, we have seen similar adoption lags. Despite continually increasing digital marketing investment, marketing effectiveness declined between 2005 and 2016. That’s a full decade of reduced productivity while marketing teams experimented with a constantly evolving digital landscape. Technology typically takes a decade or more to reach mainstream UK adoption, with the US taking around 5 years due to looser regulations and labour laws, and Europe lagging the UK some time later.
Implications for AI Adoption
Looking at historical patterns, we can anticipate AI’s future trajectory and prepare accordingly. There are three areas to focus on:
Measurement
Productivity shouldn’t be the sole metric of success. The industry’s focus on AI to reduce headcount through cost-cutting is short-sighted. Klarna’s failed attempt to replace customer service agents with AI demonstrated that AI still cannot handle complex, nuanced, or emotional issues. When these skills are needed, people prefer to talk to people.
While early adopters may achieve short-term cost improvements, this inevitably leads to price erosion. The opportunity lies in using AI to deliver higher-quality products and services that sustain price premiums. Agencies must leverage AI alongside humans, not as a replacement, creating space for humanity rather than imitating it. The focus should return to what made advertising valuable: Creativity, human insight, and ideas, not mechanical production.
Adoption Strategy
Despite the hype, AI adoption is struggling. One-third of UK companies claim to have adopted AI, yet MIT research shows 95% of AI projects failed to generate returns. Failures stem from unclear outcomes, poor data quality, skills gaps, overambitious scope, and critically, failure to integrate AI within business processes and systems. This creates an enormous opportunity for agencies through two approaches:
The incrementalist bottom-up approach involves eliminating small tasks or adding new features within existing processes. Success requires a thorough understanding of business processes, identifying granular tasks, and asking whether AI could improve cost, speed, quality, or value. Implementation must be controlled, manageable, and clearly measured—a multi-year continual improvement process.
The transformational top-down approach recognises AI’s broad impact shifts value sources in business processes and models. Businesses that rely on vendors to automate tasks face existential risks. Agencies must regularly assess what business they’re really in and where value originates, considering how AI changes not just their business but also how it impacts customers, suppliers, and competitors. Low regulatory pressure and high task-oriented job concentration make media and creative industries particularly exposed, but they are also positioned to adapt first and share lessons with clients facing less immediate pressure.
Pricing Models
A shift in value demands a change in pricing models, not just price points. Agencies face a double threat: Investing in AI while using it to discount services through reduced billable hours. However, AI’s infinite scalability creates an opportunity to profitably serve previously priced-out buyers. There are two strategies that can counter commoditisation:
Change pricing models to reflect that AI decouples service charging from the effort required
Move to different value chain positions, i.e. charge for other services which are more valuable.
The last 20 years have seen marketing increasingly shift toward operational tasks. “Marketing Operations” emerged as the fastest-growing marketing job title post-2005. Salesforce reports that marketers spend 60-70% of their time on tactical execution, while McKinsey shows that CMOs spend 70% of their time on operational management. The issue is that AI threatens the operational tasks that now make up most of marketing’s workload. This means value will shift from operational tasks such as content production, media planning, personalisation, A/B testing, and reporting, towards strategic activities such as brand positioning, customer insights, creative thinking, strategy, and decision-making.
There is time to act, but don’t waste it
Historical technology waves have shown that agencies do have time to adapt to AI, but that time must be spent actively planning for change. This means reimagining business models, identifying new sources of value, redesigning workflows, and establishing new success metrics. Agencies that learn through experimentation will become advisors to clients on similar journeys. However, simply waiting for existing vendor tools to add AI features is profoundly risky. Much of the value in new technology is captured by tech vendors that disrupt others’ business models. Expect companies like Adobe, Google and Salesforce to take value, not enable it. Now is the time to proactively rebalance services, pricing models, and operations before disruption forces reactive crisis management.
