Customer-Centric AI at Scale: How Dunnhumby Engineers Retail Data for Real Impact

What does it really take to operate customer-level analytics at global retail scale — and where does AI genuinely add value? In this episode of The esynergy Podcast, Ulrike Eder sits down with Ben Burdsall, CTO of Dunnhumby, to explore how one of the world’s leading retail data science organisations puts the customer at the centre of every decision. 

Dunnhumby operates within a complex three-sided marketplace connecting retailers, consumers, and CPG brands — balancing loyalty, shelf performance, and commercial tensions. Ben explains how tracking 2.1 billion individual customer profiles (and processing roughly 24 billion records weekly) requires a fundamentally different approach to data architecture, performance, and AI adoption. 

From building proprietary C++ database accelerators to reduce reporting times from hours to minutes, to balancing experimentation with operational stability, Ben shares what it takes to engineer for scale — and why innovation must start with the problem, not the technology. 

Key topics covered: 

• Operating customer-level analytics at massive retail scale 
• Why Dunnhumby built proprietary database accelerators in the cloud 
• Balancing traditional AI and generative AI (“staples vs cupcake”) 
• Personalisation done responsibly — and the risks of getting it wrong 
• “AI as a judge” and managing generative AI reliability 
• Bridging batch retail processing with fast, interactive AI experiences 
• Data ethics, privacy, and delivering genuine customer value 

This episode is essential listening for CTOs, data leaders, retail technologists, and AI strategists navigating scale, performance, and responsible AI in complex commercial ecosystems. 

Available on Spotify, Apple Podcasts, Amazon Music, and YouTube.