[Interview] Agata Adamczak, Head of Organic & AI Performance, Havas Media Network

As AI agents reshape how consumers search, compare and buy products, brands must rethink how they stay visible and earn trust.

Agata Adamczak, Head of Organic & AI Performance at Havas Media Network, explains how AI-driven discovery is changing marketing, why reputation and accurate information matter, and how brands can connect AI visibility to measurable business outcomes.

What is the role of the new Organic & AI Performance practice?

We help brands stay discoverable as people ask AI to research, compare and act for them.

For example, someone might ask an assistant to find a coffee machine under $700 that fits a small kitchen, makes good flat whites and can arrive before the weekend. Ranking for “coffee machine” is only part of the decision; dimensions, performance, availability, delivery and reviews also matter.

Data intelligence shows where brands appear and which sources inform answers. Content strategy identifies the evidence customers need, while technical optimisation helps search engines and AI systems access it. Consulting translates findings into actionable priorities.

Within CSA and the wider Havas Village, we bring together content, PR, media, ecommerce and customer experience teams to implement them.

How is AI changing the way consumers search for and discover brands?

People are moving from asking for information to handing over a task.

September's launches of Meta's Muse and OpenAI's Dots illustrate this shift. Muse supports everyday tasks such as travel planning and shopping, including through WhatsApp, while Dots can work across connected apps and continue delegated tasks in the background. Availability remains limited by market and plan, but using an agent is becoming more like asking someone for help.

Voice search adds another dimension. Meta has announced plans to bring Muse to its AI glasses, pointing towards journeys where people can ask, listen and decide without opening a browser or looking at a screen.

Imagine a parent asking for a nearby restaurant suitable for children with a table available at six. An assistant could compare options and recommend one for approval. This convenience could make agentic AI accessible to people who would never build a workflow themselves.

You've been clear that "AI Visibility cannot simply be treated as the next version of SEO." What's the fundamental difference in mindset that brands and marketers need to adopt?

Brands must think beyond their own websites. Assistants may encounter businesses through booking platforms, reviews, journalism and expert commentary. Reputation is therefore central to AI visibility because what others say can influence recommendations.

SEO remains foundational, but PR, customer experience and product teams also shape the evidence available. A new acronym does not remove the need to earn people's trust.

How does AI visibility translate into business outcomes like conversions and revenue?

We need to measure the journey beyond human website visits.

As assistants answer questions and compare options, some journeys may generate fewer human visits while machine requests become more important. Falling sessions alone cannot tell us whether demand has disappeared, moved elsewhere or gone to a competitor.

We examine three things together: brand presence in relevant AI answers; human visits and conversions, including identifiable AI referrals; and machine access to information. Server logs can show which identifiable crawlers or agents request product pages, whether they receive usable responses and where they are blocked.

Training crawlers, search crawlers and assistants retrieving pages for users serve different purposes, so we must distinguish them. The goal is to connect access and recommendations to business outcomes while acknowledging attribution gaps. Machine traffic is a diagnostic signal; revenue and customer value remain the outcomes.

How can brands manage AI misrepresentation risks?

Inaccurate answers can cost brands sales or direct customers towards the wrong purchases. An assistant might exclude a hotel because an outdated listing says it does not accommodate children, or recommend an unavailable product because its feed is stale.

We investigate recurring errors, examine visible sources and help brands correct information across their websites, product feeds, retailer listings and relevant third parties. Brands cannot control every answer, but maintaining accurate information must become an ongoing operational responsibility.

How will AI-driven discovery evolve, and how can brands prepare?

I expect discovery, comparison and checkout to become a more connected conversation, with agents handling more of the work.

Coles is working with OpenAI to explore personalised planning, shopping and checkout. Woolworths' Olive helps customers plan meals and add suggested products to their baskets in its app, although customers still complete checkout themselves.

For grocery brands, selection may increasingly depend on whether an assistant can match products to household budgets, dietary needs and meal plans.

Google's Universal Commerce Protocol and OpenAI's Agentic Commerce Protocol are developing connections between assistants, retailers and commerce systems. OpenAI's move towards product discovery and merchant checkout shows how retailers can retain transactions while AI helps customers decide.

Brands should test whether authorised assistants can reliably move from product discovery to completed transactions, with clear permissions and a route back to a person when needed.

Additional thoughts?

My concern is that the industry responds with more dashboards while leaving the customer journey untouched.

As average content becomes easier to produce, original research, first-hand expertise and genuine customer experiences become more valuable. Brands must give people and the systems assisting them something worth referencing.

The next visitor to your website may be a machine representing a person. Our job is to help brands serve that person's needs, even when they never see the page.

www.havasmedianetwork.com

Follow Marketing Report on LinkedIn