AI is paying off but only for those who keep their nerve

Brand owners and their agencies may view AI as the “frenemy” – insisting they cannot survive without the tech even though it threatens their very existence – but those professionals who remain focused on long-term goals are the ones most likely to see the best results.

This is according to the AI in Marketing: Leadership and Impact in Europe report from Publicis-owned data giant Epsilon, which quizzed 300 senior level marketers in Europe, across the UK, France and Germany.

The report provides insights from marketing leaders in consumer packaged goods and retail to assess AI adoption, levels of investment and measurable impact, exploring how AI tools are reshaping key functions, as well as the technical challenges and internal barriers that brands face.

It reveals that, while AI is transformative, it is clear organisations are at very different stages in their AI adoption.

Only 16% of marketers using AI see themselves as “leading”, meaning they are developing their own in-house models, building proprietary tools or using agentic AI for full campaign management.

Across Europe, France is just ahead with 19% of marketers at this more advanced stage, followed by Germany at 17%, and the UK at 13%.

Marketers report their organisations are most commonly in the “scaling” phase (36%) where AI is being integrated across multiple channels to support media buying and personalised content.

At the more tentative end of the spectrum, 21% are still in the “exploring” phase, testing tools like ChatGPT and AI-powered assistants for basic tasks. The UK (27%) is more likely to be experimenting, compared to France (22%) and Germany (15%).

Those in the exploring phase tend to have lower expectations in terms of the benefits of AI than marketers further advanced. Just half in this phase expect to see revenue growth from adoption – 10% lower than other, experienced respondents.

Their expectations about reducing media ad waste are also slightly lower than those organisations in the other phases – 31% compared to 34%.

The report flags up a number of other issues, including barriers to adoption, with concerns over data privacy (37%) still looming large. There are also fears over high implementation costs, difficulty integrating AI tools (33%) and limited internal expertise/skills (30%).

Further down the list, lack of stakeholder buy-in (20%), a lack of measurable ROI (20%) and resistance to change (19%) are less of a concern but nevertheless still an issue.

Meantime, while the vast majority of marketers are aware of upcoming regulations, such as the EU AI Act which comes into effect in August 2026, only 28% have a clear compliance strategy in place with dedicated resource; half are developing a response, but around a fifth (21%) have no concrete plans in place.

Finally, just 6% of senior marketers say it is unlikely their organisations will allow AI to make autonomous decisions. Already 47% of “leading” organisations allow AI to make some decisions autonomously, while only 18% insist on full human oversight.

This, the report maintains, reflects a key realisation that, for AI to scale effectively, it must be given the room to act – not just observe.

Epsilon director of platform solutions  Esme Robinson said: “What stands out is how expectations evolve once AI tools are actually deployed. They deliver on topline metrics like revenue and acquisition, but gaps emerge when it comes to deeper performance indicators like conversion or return on advertising spend.

“But without a connected view of the customer across channels, AI doesn’t know what it’s optimising for – you lose precision, and with it, efficiency. The marketers seeing the strongest returns are the ones anchoring AI in rich, first-party data and a clear customer identity.”

Related stories
Marketers jump on AI with gusto but keep a tight grip
Doom to boom: GenAI ‘boosts revenue, jobs and pay’
Frenemy at the gates: Industry battles with future of AI
Poor access to data is forcing bosses to turn to gut-feel
Come clean: ‘Dirty data’ costs UK firms £900bn a year
Get data shipshape for GenAI or risk sinking, firms told
Get to grips with data hygiene if you want to clean up