Dr. Jonathan Mall
ABOUT THE TALK
Synthetic shoppers on trial: a Volkswagen x Kantar head-to-head and the three ways digital twins can make and break your offer success
Every e-commerce team ships product pages, ads and claims on gut feel, because a real pre-test costs weeks and five figures. In 2026 we started replacing that step with Digital Twins: AI personas built from survey profiles that promised to answer the way the shoppers behind them would.
This session is the evidence, not a pitch. Volkswagen ran three executions of a consideration-boosting asset through the twins and through a Kantar panel in parallel: same ranking, same winner. dm's Markenspiegel used twins to rate 113 German brands across four generations, validated against real survey data first. Two enterprise validations landed at 92% and 98% overlap with human answers.
You will also see what did not work. Numeric ratings herd, so we switched to verbal-first questioning. A 100% twin consensus is a direction signal, not a measurement. And there are asset types where synthetic shoppers still fail.
Attendees leave with a decision rule for when to trust a synthetic pre-test, the 24-hour setup for their own product pages and campaigns, and the three mistakes that make the method look better than it is.