Fujitsu and FamilyMart will use causal AI to investigate why shoppers buy products with a lower carbon impact, starting with FamilyMart’s Convenience Wear clothing range.
Announced on 20 November, the project runs from November 2026 to February 2027. Marketing tests are planned for January at selected Famima!! stores in Japan.
This is a limited purchasing-behaviour trial, rather than a nationwide rollout. It begins with clothing, not groceries, and neither company has published results demonstrating a sales increase.
The partners will combine sales data with questionnaires delivered through FamilyMart’s Famipay app. The research includes people who buy the products and those who do not.
Fujitsu’s Kozuchi causal AI and large language model technology will help identify factors that may influence a purchase. The planned tests will examine how environmental information and benefits for the individual shopper affect decisions.
The initial products include Convenience Wear innerwear made with recycled polyester. Product information will provide the starting point for testing how different messages perform in stores.
For Japan’s retail technology suppliers, the project addresses a familiar commercial problem: stated interest in sustainability does not necessarily explain what reaches the till.
A questionnaire can reveal what shoppers say matters. Transaction data shows what they bought. Testing a change in communication is the next step towards understanding whether the message itself influences that behaviour.
That makes the project different from a straightforward sales forecast. Its purpose is to examine possible causes of purchasing decisions, rather than simply identify products that sell together.
It also sits alongside Fujitsu’s wider retail AI trials, while addressing a specific question about lower-carbon products and customer communication.
The partners envisage applying the findings to food and daily necessities. That remains a potential extension; the current announcement does not confirm a food-category launch or a timetable for one.
Grocery buyers and brands will need evidence of measured purchasing changes before drawing wider conclusions. January’s store tests will be the first practical check of whether the approach can turn environmental information into a different buying decision.








