Dentsu’s incoming data, tech chief discusses how generative AI could shape the future

“Now we’re in an environment, thanks to generative AI, where all of that data is at your fingertips pretty much instantaneously,” said Zelcer, who is a bit of a rarity in the holding company world as a female head of tech. “I think that’s a huge differentiator for us right now — giving our clients access to that kind of information at their fingertips.”

In March, Zelcer transitioned from her role as global head of analytics and technology at Merkle, Dentsu’s customer experience management company, where she oversaw a division of 5,600 staffers across different regions. Zelcer now reports to Michael Komasinski, CEO of Dentsu Americas and global president of data and technology, and leads a newly-formed team to focus on client-facing technology — which includes data products like Dentsu Connect and Merkury and AI across their core practice areas of media, CXM and creative.

In this interview, Zelcer discusses how data practices will change the roadmap of AI, the agency’s approach to tackling third-party cookie deprecation and why bias and privacy will become the biggest issues to tackle in the development of generative AI.

Tell me about your new team.

The team is essentially analytics, insights, products and platforms. [Within] Global Services and the way that’s all coming together in my mind is that we are going to really be the area of Dentsu that takes in the data, all our identity capabilities, the core heritage that we have and enables everything at Dentsu through it. Essentially my group is in service of everything that we do at Dentsu, whether it’s creative or media or experience and CXM — we are the enablers of that through data and tech and AI.

How does Dentsu differentiate itself in the AI competition?

The fact that we are rooted in data and we’ve been working with data for decades and decades, it gave us a different perspective on how to use generative AI. Generative AI is fantastic for a lot of the language models that we’re seeing and a lot of the chatbots, [as well as] creating website experiences and for creative and copy — but we also wanted to think about it from a data perspective and how it can kind of create ease in the data world.

How are you growing client services?

A lot of what we have built and put in place and given access to our clients is, how do we take all this great generative AI and create insights in real-time and recommendations off of those insights and generate audiences that can then be activated? That’s really where my focus has been over the last year, really enabling data and data access and insights and audiences and leveraging all of the power of AI for that.

How are you tackling cookie deprecation?

[We are] having clients understand the importance of their first-party data and what they can do with it … and making sure that our clients are prepared and getting themselves future ready, as we like to call it. Our product Merkle GenCX is built in the client environment. It takes and leverages all of their data, whether it’s first-party data, third-party data, external data that they want brought in — anything of their choosing in their environment. It allows our clients to be able to generate insights and get recommendations on those insights in very rapid ways, which they were not able to do before. 

Also, AI has been a huge momentum changer, and it’s funny because obviously we’ve been using AI for a long time – but specifically, the generative AI that has come up in the last year or so has really shifted and evolved the way that clients are thinking.

Do you have reservations and concerns in emerging technologies?

There’s a lot to think about and a lot of considerations around privacy and security, [and] around ensuring that you’re getting the right data and the right responses from your API. A lot of the challenge with generative AI specifically is that it could be generating something that you don’t necessarily want to put out there. So how do we create guardrails? How do we ensure that we are constantly monitoring what we’re putting out there? There’s a lot of bias that exists in AI, which essentially learns from the world — and unfortunately the world is full of bias. We’ve been spending a lot of time figuring that out and making sure that the models that we generate are taking that into account.