Traditional research tells you what consumers declare but rarely about the context of those declarations, particularly on shifts. Meanwhile, a large, fast-moving signal sits untapped: what consumers say and do across social media and search behaviour, most importantly how those behaviours changed over time.

In this fireside chat, Narek Garit, Global Commercial Analytics Manager at The HEINEKEN Company, shares how to integrate real-time consumer voice into an internal conversational AI platform to give deeper insights to marketers quickly and in plain language. The goal is agility, equipping marketers to react to trends and shifts in sentiment and to understand the "why" behind the numbers.

This session will provide a practical walk through of Narek's approach at HEINEKEN. We'll explore why "rubbish in, rubbish out" makes content formatting and guardrails more important than the model itself and why the smartest teams reverse-engineer from the business objective rather than building a vast data lake and hoping for value. Narek will also share his approach to measuring consumer voice, making the case that, much like a focus group, its real worth lies in qualitative understanding rather than a manufactured ROI figure.

Attendees will learn:

• How to complement survey data with real-time "say and do" signals to uncover the "why" behind consumer behaviour and democratize insights.
• How to make an LLM useful for insights, including formatting content for retrieval, building guardrails to mitigate hallucinations, and keeping customer data secure.
• Why consumer voice should often be detached from hard ROI calculations, and how a focus-group mindset leads to more honest decision-making.