What's really on the mind of digital marketers?
What we heard on the ground at DMEXCO
Elisabeth Oruba, Principal Digital Strategist
Agentic automation ran through almost every conversation I had. But the question has moved. Instead of asking what AI could do, people now ask how to get it working inside the marketing environment they already have. Embedding it with their content, their systems and their workflows.
The fear of being behind is still everywhere. However, there is little evidence for it. Almost everyone I spoke to was at much the same stage, usually because compliance, security, and EU regulatory requirements held things up in IT long before anyone got to the interesting part.
Which is why the hard problem now is adoption, not access. Getting the tools through the gates was the old challenge. Getting them into how people work every day is the current one, and that is change management, not procurement. It changes processes, responsibilities, and skills.
There is a trap in that. Scale AI across teams with very different levels of literacy, inconsistent instructions, and vague ownership, and you do not just scale the efficiencies. You scale the weaknesses too. Call it AI adoption debt.
So leaders have two jobs now, and they pull in opposite directions. Set the guardrails and the standards. And give people the confidence to use the thing.
Szymon Heliosz, Principal Experience Designer & Strategist
Marketers at DMEXCO are tired of the AI label, and the booths made it easy to see why. Nearly every stand had an agentic story, so the claim on its own told me very little.
What did tell me something was a live demo. I showed GEO optimisation agents working inside the CMS, and one question came back again and again. Can the agent actually act in the CMS, or does it just give me feedback I then have to implement myself?
That question is the line between a smarter assistant and a real change in how content teams work. But answering it changed what I think is actually on marketers' minds.
Once an agent can act, the next question is control. What is it allowed to change, and who stays in charge? Capability is not the hard part any more. Delegation is.
So here is what I would ask of any agentic offering, including ours.
Can it act, or only advise? An agent that hands you a to-do list is a reporting tool with better branding.
What does it ground its work on? Your own content, data and rules earn trust, while generic knowledge produces generic output.
And who approves it, and can it be undone? Scoped permissions, a clear review step, and an easy way back are what make it safe to let an agent act at all.
AI fatigue fades when the answers are concrete. The teams getting the most out of agents start with one specific job to be done, and design the human role around it from day one.
Greg Kalucki, Commercial Director
While the whole place was abuzz with talk of agentic AI, I found the questions that kept coming up were about marketing automation.
Specifically, B2C firms sending email at volume, where deliverability decides whether any of it works at all. And the biggest struggle with that, compliance. GDPR and EU-based data storage came up over and over, as the deciding factor in which marketing platform to choose in the first place. Several people wanted to know how Optimizely compares with other popular automation platforms on exactly that point.
Around those were the kind of conversations that tell you where people are. A retailer asking how to keep customer loyalty intact while changing its loyalty programme. A brand trying to hit consistent KPIs across its channels, so the data it already collects turns into something it can act on. A fast-growing company running marketing on basic WordPress setups, aware their stack needs to grow up, but unsure what the next step looks like.
I could hear other Igniters having conversations around AI, but mine tended to start with what needs to work well before you implement it.
Florian Westermeier, DACH Client Partner
The question I heard most was how much AI a company can realistically use in its content and distribution. Many had been disappointed with their results so far, and some had even pulled back because their audience rejected it outright.
In construction especially, where the customer base is conservative, using AI in advertising and content drew a hard pushback. So those marketers are moving people back into content creation and keeping AI for support rather than for the finished thing.
At the same time, everyone wanted to talk about LLMs as the place their customers are heading. Someone from an insurance company put it most sharply. She wanted to know how her future customers will choose an insurer once an LLM is doing the choosing, what happens when an agent compares on price and switches, and what her website and her content can realistically do to influence any of it.
Others asked a broader version of the same thing. How will people find our products, and how do we get our brand ready to be read by a machine?
Which is an awkward place to stand. Your audience does not want AI writing your content. But more and more, AI is what actually reads it.
What we took home
The hard questions were about trust, ownership, and the pace a real organisation can take. Those you can start on now, without waiting for the next model release.
If you would rather see it running, we can walk your team through the same demo we took to Cologne. Get in touch.
- Grzegorz Kałucki
- Commercial Director Poland
- (+48)664025979|grzegorz.kalucki@noaignite.com
