When AI Knows Too Much: Personalisation That Builds Trust Vs Breaks It

Sam Bradley brings a trust-first playbook shaped by years of work across Expedia, Stitch Fix and PayPal, where the cost of “getting it wrong” shows up as hesitation, drop off and long-term brand damage. His view: trust needs restraint. Minimise data use, add hard guardrails, and treat “creepy” as a measurable risk.
Pallavi Modi (zooplus) takes the opposing stance: trust needs context. Personalisation does not fail because it is “too much data”, it fails because it is the wrong data at the wrong moment. When teams confuse personalisation with segmentation, they start guessing intent (discounts, nudges, offers) instead of understanding context, journey stage and real-time signals. The result is not just lower conversion, it feels like “you don’t know me”.
Together, they debate where personalisation genuinely helps people and where it quietly becomes manipulation, especially when money and identity are involved.
Outcomes

• Put trust metrics alongside conversion so you can spot erosion before it hits revenue.
• Separate segmentation from true context-driven personalisation, and diagnose when “relevance” becomes assumption.
• Set clear guardrails for AI recommendations, offers, and targeting in high-trust, money-adjacent journeys.
• Align product, growth, legal, marketing and research on who owns personalisation, and what “too far” actually means.