Previously, generative artificial intelligence was confined to isolated test environments where enterprise software yielded only marginal productivity gains, but it now demonstrates measurable financial return inside traditional financial institutions. A pilot deployment of ChatGPT Enterprise at Credit Saison delivered a 954% return on investment. Credit Saison tested ChatGPT Enterprise with 315 employees over slightly more than a month, from late June to early August 2025. This accounted for nearly 10% of the company's approximately 3,700 employees in Japan.
Chief financial officers evaluating deployment costs against operating margins face persistent data gaps, but structured internal reporting now bridges that deficit. Credit Saison reports a 954% ROI and a savings of 14 hours per employee monthly, equivalent to roughly 169 hours annually. Both metrics significantly exceeded the company's initial targets of a 500% return and 7.5 saved hours per person per month. In a post-pilot survey, 75% of executives and 70% of frontline employees stated they could no longer imagine working without ChatGPT Enterprise.
Strategic context
Traditional financial institutions face friction when scaling local AI tests to thousands of employees without losing managerial control or incurring runaway inference expenses. Credit Saison's core business relies on credit cards, serving roughly 31.16 million cardholders. Overcoming organizational inertia required direct involvement from senior leadership from the earliest testing phases.
The pilot group brought together volunteers, employees nominated by the project team, all top executives including the president, and representatives from every department. Kazutoshi Ono, Senior Managing Executive Officer, CDO, and CTO of Credit Saison, set a clear direction by stating that the organization could not evaluate the technology's utility without direct practical application. Executive engagement went beyond passive interest. Ono encouraged Katsumi Mizuno, Representative Director, Executive President, and COO of Credit Saison, to lead by example, noting that it is insufficient to simply urge employees to use a tool—leaders must apply it regularly themselves. Mizuno reinforced this approach by building his own custom GPT for corporate research and integrating it into daily workflows.
Executives must use it regularly themselves.
Executives tested the AI independently while employees identified practical use cases. Together, these efforts tied investment decisions to daily operations, enabling Credit Saison to approve a company-wide rollout just one month after the pilot concluded. In September 2025, Credit Saison provided access to ChatGPT Enterprise to all employees.
What came before
While generative software previously lingered in pilots with negligible productivity gains, the Japanese card issuer now reports a 954% return on investment. According to the company, the test group saved 14 hours per employee each month, prompting a swift company-wide rollout. Hands-on involvement from senior executives like Katsumi Mizuno helped overcome internal inertia, though this early momentum must still survive everyday routine across thousands of desks. Will the reported efficiency hold once the tool moves past the motivated pilot cohort to the entire staff at Credit Saison?