The hard stats on where the industry stands with AI

Through its 'State of AI in Advertising' report, the IAB UK has been examining how advertisers, agencies and consumers are adapting to that shift. Combining interviews and quantitative surveys across the advertising ecosystem, the research explores where AI adoption is accelerating, where structural pressures are beginning to surface and what may ultimately determine whether the industry can scale the technology sustainably.

The findings focus on the rooms seeing the most disruption: adoption, trust, content creation, zero-click search, governance, sustainability and agency value.

We spoke to Jeremy Pounder, AI Consultant, and Alex Kozloff, Director of Industry Relations, to unpack what the findings reveal about advertising’s AI transition in 2026.

Jeremy Pounder:

The three biggest factors are: data foundations, organisational complexity, and culture.

Data foundations: how companies capture, structure and access data is a major factor in how ready they are to take advantage of AI tooling.

Organisation complexity: Some of the dimensions of the report showed, counterintuitively, that smaller advertisers are in some ways more advanced than larger ones. You tend to assume bigger equals more sophisticated, but sometimes bigger also means more legacy systems, more bureaucracy, and more silos in how data and processes are set up.

Culture: Some organisations are fundamentally more progressive, and that’s potentially linked to how well organised they are from a data and digital perspective more generally.

Overall, we were surprised at how ready people felt. We expected more advertisers and members to feel anxious about the scale of transformation AI is likely to bring.

From an industry perspective, this a positive sign as I’m not sure we would have seen the same results a year ago.

Jeremy Pounder:

With some of these agentic systems being stitched together, there is definitely a risk of unknown consequences that are difficult to fully anticipate.

The digital advertising industry is already very complex and has historically faced challenges around transparency. Introducing agents that make decisions based on probabilistic rather than deterministic technology adds another layer of uncertainty into the system.

That becomes even more pronounced when you’re stitching together multiple decisions and multiple agents, where the decision-making process is not always explainable to people. So it’s understandable that this is reflected in the trust statistics from advertisers.

Alex Kozloff:

While it’s surprising to see how quickly the industry is moving, we’ve found that advertising and digital advertising are actually further ahead than many other industries when it comes to adoption. 95% of businesses in the sector are now using the technology, while just 16% of UK businesses overall report using AI.

There is clearly anxiety around AI, both within society and within the industry itself, but advertising has always been a fast-moving sector and historically a high adopter of new technology.

Jeremy Pounder:

Ultimately, it shows that we are still in the relatively early stages of adoption.

The low percentage wasn’t particularly surprising because “agent-first” sets a very high bar. There are probably very few organisations across the wider economy, let alone within our industry, that could genuinely describe themselves that way.

Agentic AI as a concept is still very new. Within our industry, it’s probably only been part of the conversation for 12 to 18 months. Many of the standards, governance and operational structures needed are still being formed.

For established businesses especially, it would be difficult to move from where they were 18 months ago to becoming genuinely “agent-first” in the way we defined it. Reaching that level of integration probably lends itself more naturally to startups or companies building from scratch.

Beyond the major tech platforms that are defining the market, it may actually be medium-sized companies that find the agentic AI transition hardest. Smaller businesses are often more agile and can start with a cleaner slate when adapting to an agent-first world.

Jeremy Pounder:

Businesses are naturally going to feel more confident scaling investment in generative AI systems if they believe the right guardrails and governance protocols are in place.

That could mean preventing ad creative from appearing in the wrong environments or avoiding reputational and brand safety risks.

Agencies are also very conscious of governance because of their role acting on behalf of clients. One of the concerns we picked up, particularly from media agencies, was around sell-side agents having too much autonomy in executing buying decisions.

There is a strong emphasis on governance procedures that allow decisions to be audited properly and ensure accountability around where money is being spent.

Jeremy Pounder:

Consumers are increasingly aware of the wider phenomenon of “AI slop” and the volume of low-quality AI-generated content online.

That said, most of what consumers see that is AI-generated probably sits outside conventional advertising in the broader creative ecosystem online with social and wider internet content generally. So while there may be a broader perception that online content quality has eroded, consumers are not necessarily blaming advertising specifically.

More broadly, the media landscape has fundamentally changed over the last 20 years. The industry often looks back nostalgically at a time when TV advertising was part of the national conversation, but that fragmentation isn’t really an AI story - it’s a result of broader changes in media consumption.

From a creative perspective, I also don’t think we’ve yet seen the full impact of AI in advertising. As we outlined in the report, most current use cases involve adapting existing creative assets rather than generating entirely new campaigns or ideas from scratch. A lot of those adaptations are invisible to consumers.

Jeremy Pounder:

Consumers are now exposed to a much broader range of advertisers, particularly SMEs and smaller brands, because platforms like Meta and Google have dramatically lowered barriers to entry.

That potentially changes the overall quality of the landscape because consumers are exposed to a much longer tail of advertising than they would have been historically.

There’s also a longer-term creative risk that the industry is beginning to think about. Many AI advertising tools are effectively codifying platform best practice - things like logo placement, pacing, messaging structure and optimisation rules.

If every advertiser follows the same optimisation rules and systems trained on the same definitions of best practice, creative output could become more homogenised. This may make it harder for brands to produce genuinely distinctive creative work.

Jeremy Pounder:

AI is overhauling the economics of creative production because the barriers to producing creative assets are far lower than they used to be. The ability to personalise content and create thousands of variations is rapidly becoming much more accessible and cost-efficient.

Whether it changes creativity itself is a slightly different question. It depends on how heavily brands rely on these tools as the starting point for campaign development, and that’s going to vary significantly across the market.

Larger, more established brands are still likely to place significant value on the idea of brand essence, big creative ideas and distinctive storytelling. Those brands will probably continue to adhere to more traditional creative principles and still see agencies as central to that process.

The bigger risk is further down the long tail of advertisers, particularly among SMEs. The priority may simply be creating advertising that reaches the right people efficiently while keeping production costs low.

In those cases, efficiency and performance may become more important than originality or creative distinctiveness, and that’s where you potentially start to see more homogenisation in the market.

Jeremy Pounder:

It depends on the scale of the shift and how quickly consumer behaviour changes.

There is already evidence that traffic is falling significantly in certain sectors, particularly around news publishing and more practical content categories like food, lifestyle and informational search content. Smaller or niche publishers producing generic content could be affected much more severely if AI summaries continue reducing the need for users to click through.

For more premium or specialist publishers, the situation may be different. Some may choose to make strategic decisions around how accessible their content is to AI crawlers, while others are probably established enough to remain destination platforms in their own right.

This issue is also closely tied to the ongoing debates around copyright, licensing and compensation between publishers and AI companies. There are already commercial agreements emerging between some of the largest publishers and major AI platforms, but how that translates to the much broader long tail of publishers is still very unclear.

Jeremy Pounder:

Most practitioners seem to view GEO as an evolution of SEO rather than something entirely separate.

There’s still a strong technical layer around how websites are structured and how readable they are for LLMs.

That includes things like “schema markup” and the way information is organised on a page to make it easier for LLMs to interpret and surface content. Clearly summarising key information at the top of a page also increases the likelihood that content will be surfaced.

Beyond the technical side, there’s also an evolution of “digital PR”. Brands are increasingly thinking about how they appear across platforms that LLMs frequently draw from and how to strengthen their visibility and authority in those environments.

There’s also growing discussion around making product and first-party data more accessible to AI systems. That could include product information, customer reviews, return rates and other forms of proprietary consumer insight that retailers hold themselves.

Jeremy Pounder:

There are really two dimensions to the skills gap. 

1. The first is the need for deeper technical expertise - data science skills and the people capable of building and managing more advanced systems, including agentic AI systems. Those are highly specialised capabilities, and historically most agencies haven’t had large amounts of that talent in-house.
   
   From everything I’ve seen, there’s also a limit to how far businesses can get through lightweight experimentation or “vibe coding”. At a certain point, organisations still need genuinely deep technical practitioners in these areas.
   
   The challenge is that bringing in that kind of specialist talent is difficult given the pressure many agencies are already under commercially.
   
2. The second dimension is broader organisational adaptation. This applies to almost every business, not just agencies. It’s about understanding workflows and team structures, then figuring out where AI can genuinely improve or reshape them rather than replace them entirely. That requires training, support and a willingness to rethink how work gets done.

Alex Kozloff:

The industry has been through multiple periods of technological transformation before, but AI feels uniquely emotionally charged because people are aware of both the opportunity and the threat at the same time.

There’s a real tension around how agencies encourage people to embrace these tools without employees immediately assuming the technology is there to replace their jobs.

Agencies need to think carefully about communication, incentives and how they bring people along because it’s not simply a technical challenge, it’s also a cultural and organisational one.

Jeremy Pounder:

Potentially, yes. Traditionally, clients have used agencies partly for execution - production on the creative side and delivery on the media side. To varying degrees, those functions can now be automated or brought in-house, which wasn’t really an option even a few years ago.

On the other hand, clients have also always valued agencies for strategic advice and creative ideas. Those areas are likely to remain important.

The challenge is that agencies have historically made a significant amount of money from execution and production, while strategy and big creative ideas haven’t always been monetised as effectively - they’ve often been subsidised by delivery and production revenues.

That creates a structural challenge for agencies: shifting from a model based on time spent delivering work to one tied more closely to outcomes and effectiveness. This shift has been long discussed but it becomes even more pressing in an AI-driven environment.

There is still a strong advisory role for agencies in helping clients use these systems effectively because they have experience across categories and established relationships with major platforms. Even if we move towards a more agentic model where clients are increasingly using tools themselves to generate and distribute advertising, they will still need support in understanding how to use those technologies well.

Jeremy Pounder:

If we had asked the same question two or three years ago, I suspect environmental impact would have ranked much higher across the board. The shift was quite noticeable.

A few years ago there was a major push around integrating carbon calculators into media planning tools, with the expectation that clients would increasingly ask about the carbon footprint of one media plan versus another.

Three years later, I’m not sure those tools are being used as heavily as people expected. That may simply reflect the fact that advertisers are not asking those questions as much anymore, and agencies therefore are not pushing it as actively either.

Alex Kozloff:

I also think the sustainability impact of AI is still difficult to measure because it varies so much depending on the use case.

Certain forms of generative AI can obviously involve a heavy processing load, but there’s also longer-term potential for agentic systems to streamline processes and create more direct transaction pathways, which could actually reduce overall system load.

The industry still doesn’t fully understand what the net impact will be at scale. How AI is deployed matters just as much as the technology itself, and I think we are still trying to work out what those trade-offs look like.

Alex Kozloff:

The role of the industry is becoming more important as businesses try to align on standards, governance and trust frameworks at the right pace.

Through our sister organisation, the IAB Tech Lab in the US, there has been significant work on developing industry standards and technical specifications. They’re now doing similar work around agentic AI to help create more consistent foundations for the industry.

They’ve also recently announced a governance council designed to bring together agencies, sell-side players and other stakeholders to align on cross-industry standards and oversight.

In the UK, the IAB UK is doing similar work, collaborating with partners, members and other trade bodies such as the IPA and ISBA to address these shared challenges as they emerge.

A lot of this is still very nascent, so it’s an ongoing process that requires coordination across the industry.