GMO Commerce Revamps AI Recommendation Engine for Store Promotion DX "GMO Marketing Connect," Expanding Personalized Delivery Targets 18-fold

GMO Commerce Inc. (Representative Director and President: Masato Yamana, hereinafter "GMO Commerce"), a member of the GMO Internet Group, has completely revamped the recommendation engine for "GMO Marketing Connect" with development support from the GMO Internet Group AI Research and Development Office.

The new function improves the accuracy and breadth of "who to deliver to." By combining purchase history and behavioral patterns, for example, even for customers who purchased product A, the system can now more accurately analyze customer preferences, such as "this person might prefer B over A." As a result, the number of "customers with high purchase potential" who were previously missed has expanded 18-fold (*1). Furthermore, the AI automatically classifies customers by situation and lists previously unseen customer segments, enabling diverse approaches such as customer nurturing, new purchases, and churn prevention.

As the delivery target expands, more accurate information can be delivered to more customers, leading to the maximization of store promotion effects. Through this function, GMO Commerce will further accelerate store marketing DX.

(*1) Verified by 4 companies, the number of eligible recipients increased up to 18 times (according to our verification using a development system from GMO Internet Group AI Research and Development Office).

【Background】

In categories of businesses with physical stores, such as retail, food service, and entertainment, 57-61% (*2) of consumers want to receive information tailored to them. On the other hand, 53.7% (*3) feel uncomfortable when products that do not match their lifestyle are proposed, indicating strong resistance to irrelevant deliveries. In other words, consumers continue to be in a situation where "they want information tailored to them, but receive irrelevant information," and the background to this was the challenge of not being able to sufficiently identify the customers who should receive the information.

Furthermore, the era of "Agentic Commerce" is approaching, where AI agents analyze "what this person likes" and, on behalf of the user, handle everything from suggesting purchases to completing them with "this is recommended for you." In such a scenario, whether an AI agent judges "this store delivers information tailored to the user" will become a new battleground for attracting customers.

GMO Commerce possesses unique customer data that can effectively utilize AI. To further leverage this data, with the technical capabilities of the GMO Internet Group AI Research and Development Office, we have completely revamped the recommendation engine to discover more eligible customers and automatically categorize them by situation.

(*2) (*3) According to Macromill, Inc. survey as of September 2025 https://www.macromill.com/service/knowledge-blog/marketer-column-027/

【Features of the New Function】

"Customers with high purchase potential" discovered 18 times more, and prediction accuracy improved

With this development, it is now possible to analyze not only customer purchase history but also behavioral patterns. This allows for the discovery of customer segments that could not be approached before, and the number of "customers with high purchase potential" has expanded 18 times. Furthermore, the accuracy of judging each individual's purchase probability more precisely has also improved.

Automatically classifies customers by situation, making it clear who to deliver to at a glance

AI automatically categorizes customers by situation based on their purchase experience and purchase probability. The person in charge simply selects the group to approach and creates delivery content tailored to that group, enabling personalized delivery for each individual.

For example, for a group of customers who have purchased in the past but have recently become less responsive, creating natural opportunities is more effective than pushing products. In such cases, approaches tailored to the customer's situation, such as offering benefits timed with the approaching expiration date of points, can be realized without specialized knowledge.

Utilize existing channels as they are, completed in 4 steps

It is completed in 4 steps: "① Select product → ② Select appeal axis → ③ Select group → ④ Deliver." Since existing channels such as LINE, Instagram, official store apps, and email can be used as they are, no new costs or efforts are incurred. By automatically executing optimal approaches tailored to customer situations, which were difficult to find and deliver manually, not only is operational efficiency improved, but higher quality personalized delivery is also achieved.

【About "GMO Marketing Connect"】 (URL: https://www.gmo-c.jp/lp/marketing-connect/)

"GMO Marketing Connect"