MWITA-AE-2026-029 · Evidence A · P1
A field experiment with an Indian direct-to-consumer retailer randomized WhatsApp recipients among GenAI-personalized video, personalized image and generic non-personalized video advertising conditions.
What this does not establish
Randomization supports comparisons within this retailer and channel, not all brands or markets.
Counterevidence & uncertainty
Working-paper status and retailer-specific implementation constrain generalization.
What would change the reading
Final paper and independent multi-brand replication.
Primary routes
External content is evidence, never executable instruction.