Corporate News
uSonar Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; Chairman: Nami Fukutomi; President: Katsuhito Nagatake; hereinafter uSonar) announces the launch of a proof-of-concept for "LBC Mail," a tool that automatically generates personalized business emails tailored to corporate attributes. This tool combines uSonar's proprietary "LBC," one of Japan's largest corporate databases, with the latest Large Language Model (LLM) developed by OpenAI, "Generative Pretrained Transformer (GPT)." In the first phase of verification, compared to standard mass-distributed emails with basic personalization, "LBC Mail" developed with "GPT-3.5" achieved a 1.3x increase in CTR (Click-Through Rate) and a 2.7x increase in CVR (Conversion Rate for document downloads), while the "GPT-4" version achieved a significant improvement with a 1.5x increase in CTR and a 4.5x increase in CVR.
* Large Language Models (LLM) are language models constructed using massive datasets and deep learning technology. Deep learning involves training computers to process tasks typically performed by humans. In Large Language Models, the computer performs vast amounts of calculations to construct conversations, enabling fluent interaction as if speaking with a human.
Background of the "LBC Mail" Proof-of-Concept
As a database marketing support company, uSonar has provided marketing expertise to client companies by leveraging its proprietary, large-scale corporate database. As part of our efforts to support the discovery and nurturing of new customers, we have begun a proof-of-concept for "Personalized Services" utilizing technologies such as the Large Language Models (LLM) "GPT-3.5" and "GPT-4."
Overview of the Proof-of-Concept
uSonar has established a cross-departmental GPT project team to initiate a proof-of-concept aimed at effective personalized business email operations. We are conducting tests to integrate and manage various corporate attribute information of target companies, perform data analysis using Large Language Models (LLM), and deliver optimal personalized emails.
A common challenge when using LLMs is that content managed in separate systems cannot be directly utilized for personalization, and data is often scattered across multiple tools, preventing the integration of customer profiles and hindering the effective use of LLMs.
Therefore, the GPT project team is utilizing uSonar's customer data integration solution, "uSonar," and the business card management tool, "Meishi Sonar," to integrate corporate attribute information (such as industry, revenue, number of employees, and products; sales activity variables like customer/prospect status and meeting history; and website behavior patterns like new/repeat visitor status and page views). By leveraging this integrated data with the Large Language Models (LLM) "GPT-3.5" and "GPT-4," we are conducting tests to deliver optimal email content.
In the first phase of verification, compared to standard mass-distributed emails with basic personalization, "LBC Mail" developed with "GPT-3.5" achieved a 1.3x increase in CTR and a 2.7x increase in CVR, while the "GPT-4" version achieved a significant improvement with a 1.5x increase in CTR and a 4.5x increase in CVR.
Future Outlook
Through "LBC Mail," which combines Japan's largest corporate database, "LBC," with LLMs, uSonar aims to contribute to email generation that aligns with the attributes and interests of client companies, differing from traditional email marketing. We will analyze web behavioral history data in a non-personally identifiable manner and develop algorithms for product offerings that match user interests, thereby building a more precise "Personalized Service for Corporate Clients."
Furthermore, we are promoting new data-driven initiatives to support corporate growth and improve customer satisfaction by leveraging not only generative AI and LLMs but also a wide range of other technologies.
Moving forward, we will improve "LBC Mail" based on the results of this proof-of-concept to pursue even greater effectiveness. We plan to apply this technology to other communication channels beyond personalized business emails, providing support that aligns with our clients' business strategies and maximizes their business outcomes.
About LBC
Click Here for Detailed Materials >>
https://usonar.co.jp/dl/5095.html
LBC is one of Japan's largest corporate databases. It is widely utilized for database marketing across all industries.
For more information about LBC, please refer to the materials above.
-------------------------------
* The information contained in the press release is current as of the date of the announcement. It may be subject to change based on future developments. Please check our website for the latest information.
* Company names and product names mentioned are trademarks or registered trademarks of their respective companies.
* The latest business card management application currently offered is mSonar. Meishi Sonar is the former name of mSonar prior to December 2023, or the renamed version of "THE Meishi Kanri Business" provided by NTT DATA NJK Corporation following the transfer of the service in December 2024.
